analytics rfp - ve ndor q uest i ons - bidnet the platform for its own data analytics needs as...

25
Analytics RFP - Vendor Questions Q1: Information Builders has both a 2014 Master Software Licensing Agreement with UNC General Administration which covers 17 UNC campuses and a 2014 Master Services Agreement that covers 27 UNC Affiliates. Will the State agree to use Vendor’s existing agreements with UNC General Administration to govern the licensing of software and the performance of services proposed by the Vendor? A: In general, yes, but we do not make the assumption that everything in the proposal would fall into the original contract scope, which might require additional terms of service, addendums, etc. By the way, here is how we think about the structure of the existing contract, which I think is materially the same as your original description, but just wanted to clarify. Information Builders has a 2014 Master Software Licensing Agreement with UNC General Administration and the constituent institutions, as well as a 2014 Master Services Agreement with the same institutions and including nine UNC affiliates. Followup Question: Did not clearly understand the question and the clarification. A: This was a very specific question with the vendor trying to understand how an existing contract might be connected to a new proposal. We are not sure how else to clarify. Q2: What is UNC using for current statutory reporting and analysis needs out of the Data Warehouse and/or the data marts? A: We currently use a combination of a home-grown, web-based pivot table generator and SAS programming. The home-grown tool provides some simple interaction with the data for basic flat and pivot table reporting off of datasets (Oracle views); the Data & Analytics unit leverages the SAS programming language to handle various routine and ad hoc reporting needs and conduct analyses of a more complicated nature. Q3: Is UNC open to a Systems Integrator (SI) partnering with a proposed Software Vendor? A: Yes Q4: Can we assume Oracle as the database for Data Warehouse? Are there any other databases to consider, now or in the future? A: Currently the warehouse and all of the data, views, materialized views (etc.) are stored within Oracle, and we don’t anticipate this changing in the near term. If Oracle is to be the engine directly servicing the platform (in real-time), the infrastructure would need to be augmented to increase the level of concurrency and throughput to significantly scale the solution. As campuses are brought on board, they will likely want to integrate with some of their own data sources which will likely include MySQL and Microsoft SQL Server, among others. Q5: What is UNC’s current Operating System or the OS that the proposed solution needs to be compatible with? A: For an on-premise solution, we prefer Redhat Linux compatibility. Q6: How many dashboards is UNC expecting to deliver within the 4 months? 1 of 25

Upload: truongnguyet

Post on 22-Mar-2018

214 views

Category:

Documents


2 download

TRANSCRIPT

Page 1: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Analytics RFP - Vendor Questions Q1: Information Builders has both a 2014 Master Software Licensing Agreement with UNC General Administration which covers 17 UNC campuses and a 2014 Master Services Agreement that covers 27 UNC Affiliates. Will the State agree to use Vendor’s existing agreements with UNC General Administration to govern the licensing of software and the performance of services proposed by the Vendor? A: In general, yes, but we do not make the assumption that everything in the proposal would fall into the original contract scope, which might require additional terms of service, addendums, etc. By the way, here is how we think about the structure of the existing contract, which I think is materially the same as your original description, but just wanted to clarify.

Information Builders has a 2014 Master Software Licensing Agreement with UNC General Administration and the constituent institutions, as well as a 2014 Master Services Agreement with the same institutions and including nine UNC affiliates.

Followup Question: Did not clearly understand the question and the clarification. A: This was a very specific question with the vendor trying to understand how an existing contract might be connected to a new proposal. We are not sure how else to clarify.

Q2: What is UNC using for current statutory reporting and analysis needs out of the Data Warehouse and/or the data marts? A: We currently use a combination of a home-grown, web-based pivot table generator and SAS programming. The home-grown tool provides some simple interaction with the data for basic flat and pivot table reporting off of datasets (Oracle views); the Data & Analytics unit leverages the SAS programming language to handle various routine and ad hoc reporting needs and conduct analyses of a more complicated nature.

Q3: Is UNC open to a Systems Integrator (SI) partnering with a proposed Software Vendor? A: Yes

Q4: Can we assume Oracle as the database for Data Warehouse? Are there any other databases to consider, now or in the future? A: Currently the warehouse and all of the data, views, materialized views (etc.) are stored within Oracle, and we don’t anticipate this changing in the near term. If Oracle is to be the engine directly servicing the platform (in real-time), the infrastructure would need to be augmented to increase the level of concurrency and throughput to significantly scale the solution. As campuses are brought on board, they will likely want to integrate with some of their own data sources which will likely include MySQL and Microsoft SQL Server, among others.

Q5: What is UNC’s current Operating System or the OS that the proposed solution needs to be compatible with? A: For an on-premise solution, we prefer Redhat Linux compatibility.

Q6: How many dashboards is UNC expecting to deliver within the 4 months?

1 of 25

Page 2: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: We do not have a fixed number in mind to determine “success”. Our goal is to be able to make significant progress in fours months to be able to achieve quick wins and increase the confidence and excitement around the future prospects. Therefore, it is not about quantity, but about value-added. In other words, one high-value dashboard might be better than 10 average dashboards. We would like the proposals to include the vendor’s recommendations on how many high-impact dashboards would be achievable within the 4 month period.

Q7: How many Star schemas, tables, etc. should vendors plan for? A: We store the data mart information in one database schema consisting of two units, one for Students and one for HR, with approximately 200 tables. ~156 Dimensions, ~38 fact tables, once completely converted around 85 materialized view tables. The materialized view tables would be the basis of all reporting. Followup Question: Materialized view tables would be the basis of all reporting. Does it mean, we need to build ETL integration between Mat View layer of current Data-warehouse and new Analytic Layer? Should new Analytic layer consist of additional Data mart built on top of already available data mart? A: Without getting into specific technology details, we expect some level of ETL to be able to transform the materialized view tables into whatever model the analytics platform needs. So, yes, we think this is congruent with the “analytic layer” you mentioned in the question. This analytic layer would should also be able to handle other sources of data that are not part of the official materialized views.

Q8: How many initial licenses of the tool should vendors base pricing on? A: This question is hard to answer precisely due to varying licensing metrics used by vendors in this space. Therefore, we will describe the ideal situation for the initial phase. We expect to have maybe 10-12 “power users” (at UNC-GA) who would actually be able to create dashboards for the consumption of others We will have 2-3 (named) users per campus and 10-15 users at UNC-GA (so approximately 50-60) of users that we would like to be able to do some version of “ad hoc” reporting (pivots reports, data visualizations, etc.) We expect to have another category of “named users” at each campus that would be able to consume dashboards and reports created by the “powers users”, but these users would be confined to their particular institution; our best estimate would be 10-15 per campus (200-240 total). Finally, we would then have the possibility of having (anonymous, public) users who could interact with the produced dashboards; it is impossible to know how many such users we would get but we can likely assume it is below 100 concurrent users at a time. If campuses wish to join the system then the licensing would need to be expanded to allow them to add their own custom datasets, create their own dashboards, etc.

Q9: UNC mentions Travel expenses in one occasion (page 25, item #8), and Inspection of Vendor Sites in Attachment A. Travel expenses may be applicable when Vendor personnel are stationed at the UNC site to deliver the task orders. Can UNC clarify if the expectation is that the Vendor personnel should be stationed at the UNC site? Secondly, does UNC have the space to accommodate vendor personnel? A: UNC GA does not expect the vendor to be stationed at the UNC site.

Q10: The following is a requirement listed on Attachment E: “The Vendor is in sound financial condition and, if applicable, has received an unqualified audit opinion for the latest audit of its financial statements.” Smaller firms may not have audited statements, but are in sound financial condition and have financial statements available through a licensed CPA firm. Can UNC waive the audited financial statements expectation?

2 of 25

Page 3: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: Vendor can check the box regarding the “vendor is in sound financial condition” and notate on the sheet that the firm does not have audited financial statements.

Q11: Can UNC please provide editable RFP forms for completion? A: No

Q12: The RFP states that “…the selected platform and associated services would allow for all of the UNC constituent institutions to benefit from a baseline set of services and for any of the UNC constituent institutions to optionally leverage the platform for its own data analytics needs as well.” The RFP is unclear as to whether providing UNC-GA’s constituent institutions with this independent ability represents an immediate need. Should providing this ability be an immediate part of the proposal or should the response confine itself solely to providing the General Administration Office staff with modeling and reporting capabilities and simply attest to the optional extendibility of the solution? A: Although the details of timing and services provided will vary, in general vendors may want to approach this as some sort of phased implementation. We would expect that a baseline set of system-wide services (i.e., basic reporting needs) are met within the initial four-month time frame articulated in the RFP. Beyond that, we are open to alternatives and will consider the timing and specific services proposed as part of the overall evaluation of proposals. Please refer to Q8 for discussion on the types of users and functionality expected for what may be considered the initial phase of implementation. For the subsequent phase, we might expect some level of broader adoption directly by a subset of campuses that see value in the platform and wish to further invest. We do not yet know the list of such subscribers and it will partially depend on the success of the first phase and the value provided to those campuses. However, we want to understand the overall, long-term budget impact if campuses joined the solution and will consider this information into the economic viability of the overall solution.

Q13: What constitutes the “baseline set of services” that should be available to the individual campuses? A: The answer to question Q6 (above) partially answers this question. We did not specifically

Q14: Approximately how many individual dashboards are required to satisfy UNC-GA’s initial needs? A: Please refer to Q6.

Q15: How many “authorized users” are anticipated. A: Please refer to Q8.

Q16: How many “power users” should be provisioned for. A: Please refer to Q8.

Q17: Is UNC GA using an analytical tool today, and is the objective to replace the current tool? A: UNC-GA does not currently have an enterprise grade analytical/BI tool. We use an in-house developed web-based solution that allows us to do basic adhoc reporting (both flat and pivot tables). We seek to release the stock and adhoc reports currently implemented in this tool with a new tool with more capabilities. The Data & Analytics unit also uses SAS for the reporting functions that require more than simple flat and pivot table reporting. We anticipate an ecosystem of tools (all leveraging a common platform) to meet the diverse set of needs in this space.

3 of 25

Page 4: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q18: Are there requirements for integrating with any custom portal or other applications (other than Banner and PeopleSoft)? A: We are not expecting any form of “integration” with Banner or PeopleSoft. Our data warehouse already does these integrations. We anticipate the integration to be between our data warehouse (and web-based data marts) and the new analytics platform. However, we would like to be able to embed any visualizations created with the BI tool to any of our existing web-based websites/portals. Therefore, some kind of web-based API (javascript or equivalent) would be highly desired.

Q19: For predictive analytics, will UNC GA allow integration with an advanced analytics software such as R as part of the overall solution? A: Yes, that could be part of an ecosystem of solutions.

Q20: Can work be performed remotely (outside of planning and face-to-face meetings)? A: Yes, we would expect that arrangement.

Q21: Is the 4-month deadline for contents a production-state deadline? Or is UNC GA open to developing a solid prototype that can be productionized? A: See Q6 for a similar discussion.

Q22: How many users will view the system-wide aggregate dashboards? A: Please refer to Q8.

Q23: How many authorized users will review system-wide records? A: Please refer to Q8.

Q24: How many campus users are expected? A: Please refer to Q8.

Q25: How many “power users” are expected to create and publish reports? A: Please refer to Q8.

Q26: What types of local data sets will be loaded? How many users will require ability to load data sets? A: For campuses that wish to join and take advantage of the analytics platform, we would want them to be able to upload any type data they wish - e.g., student-related, finance-related, faculty information, or any other data set that they might have and would want to visualize or use in predictive models. These data would likely be in typical formats (xls, csv, sql/database, json, etc). We would expect a small number of users to have this level of access from the participating campuses; perhaps 5-10 but would certainly depending on the size of the campus and its staff. Followup Question: Would new/additional campuses be allowed to up-load any type data sets they wish? If yes, what if the existing or new model structure do not support new data attributes? A: In our vision, yes, participating campuses would be able to upload their own data sets to the platform. The platform should be sufficiently robust to be able to accept new data structures and not need to be confined by the data model of

4 of 25

Page 5: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

the official “data warehouse”. We see the “platform” as an ecosystem that can accept many different types of data and leverage different tools to access the data and transform it into actionable information.

Q27: What types of services are expected? Training on the tool? Development services? Ad-hoc or scheduled? A: We would expect training on the tool along with the consulting/development services needed to do the integrations necessary to get the platform established and into production. Since each vendor will likely have a very different approach it is difficult to provide a prescriptive statement in this area. We expect the vendor to provide recommendations on the level of services needed to be successful.

Q28: What current ETL tools are being used? A: There are currently no specific ETL commercial tools being used in the data warehouse. SQL is used within Banner/Peoplesoft to create a standard set of data feeds that are queried by a JSON-based web services framework that are eventually loaded into the data warehouse at UNC GA. There is then a set of SQL that transforms that raw data into the data model for the warehouse along with a set of materialized views that present the logic “datasets” that are used by the reporting tools. The following diagram outlines the basic architecture of the current data warehouse and how we envision it will be related to the analytics platform:

Followup Question: In data flow architecture diagram, is it a mandate to use Mat Views layer as the originating source of Analytics Platform? A: Mandate is a strong word. We are willing to entertain any proposal that makes sense from an implementation/economic perspective that also leverages the existing investment we have already made into creating the robust data warehouse. So while not totally required, it is likely such an approach will be most optimal, but again we are open to other reasonable models if the vendors make a good case.

Q29: Are there training requirements for users? A: Yes, we would expect the vendor to supply training for UNC GA staff. We would be open to a train-the-trainer situation with respect to training the campus staff, and would seek guidance and work with the vendor to determine the best approach.

Q30: Does UNC GA already have a 1-way trusted identity provider (IdP) setup internally to handle the multiple identity providers? How does UNC GA currently address this need in other applications? A: We have what we call the “UNC Identity Federation” whereby we can allow access to any central application using local campus credentials using a federated SAML approach. The Service Provider (SP) applications are “federation-aware,” so that whenever a user attempts to login they are presented with a Where Are You From (WAYF) interface that enables them to select their campus. The user is then redirected to that campus Identity Provider (IDP) and a SAML message is then returned to the SP for proper login. Here is a screenshot of a sample such WAYF integration:

5 of 25

Page 6: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Our strong preference is to the have SP federation-aware, but if this is not an option then we can consider using an IDP proxy to implement the WAYF functionality in a one-to-one SP-IDP configuration. Followup Question: Would BI application portal need to be integrated with UNC-GA intranet? What if BI application provides their own portal for reports/dashboard access? A: An external portal is an acceptable solution, but must we SSO compliant as described above. We do not want to add a user account administration burden to either our users or to UNC-GA administrators. An ideal BI solution would also allows us to embed “widgets” into existing websites via a javascript (or similar) API. See Q93.

Q31: The schedule shows both Oral Presentation and Product Demonstrations by Finalists with the date of 11/23/2016. As this is the day before Thanksgiving, do finalists need to be prepared to attend onsite meetings on Wednesday, 11/23/2016? A: This was a typo in the schedule, and it is incorrect. We have currently reserved the on-site “Product Demonstrations by Finalists” for Thursday, December 15, 2016.

Q32: Could you please elaborate on the intent of requirement 5.2.11? Is this specific to the security framework and access in the product? A: First, we want to ensure that if the solution is cloud-based, that the vendor would not have rights to mine or otherwise leverage the data for its own business purposes (such as marketing, distribution to 3rd parties, etc). Along the same lines, we would want clear understanding of liability if cloud data was in breach, how the vendor protects the data internally, etc. Furthermore, the selected platform would need to have a framework for determining how data is shared across participating institutions. For example, we would want each institution to be locked into its own campus data, but also be able to aggregate certain sets of data across all institutions. In short, we want to understand all of the details around how security and rights to the data can be provisioned, monitored, and enforced.

6 of 25

Page 7: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q33: Is there a preference for the product to be installed on Premise or Cloud? A: We are open to both models.

Q34: Is UNC preference for a complete COTS student success platform or for a custom built solution with custom dashboards and reports using a credible BI application? A: The ideal scenario would be a platform that comes with a set of (best practice, action-driven) dashboards and reports that uses a COTS BI tool that can also be used to do any custom dashboards or reports needed to address our particular business needs.

Q35: What is the required number of Environments for the Product (Dev, QA, Prod & DR)? A: We require Production, Development, and D/R; having a Test/QA would be ideal but not required.

Q36: Can you provide a list of all source systems that have data for this reporting effort? A: For the immediate phase, there is only one source system, the UNC-GA data warehouse. The underlying campus source systems are not applicable, as that complexity is masked by the implementation of the data warehouse. If campuses join the platform, then they would have alternate source systems that could be leveraged. See Q26 for discussion on the nature of these datasets. Followup Question: Is expectation to build ETL integration between Mat View layer of current Data-warehouse and new data mart? A: See the followup from Q7 and Q28.

Q37: Do all the UNC campuses have the same source systems and same version? A: Similar to Q36, this level of detail is not applicable to our current model (see Q28). However, we are open to being convinced of alternatives.

Q38: Can you provide an architectural diagram of current systems, how they’re related, and what platforms the underlying data is stored (flat files, RDBMS, Web, NoSQL etc)? A: See Q28.

Q39: Do source systems have underlying data models that we can review? A: See Q38.

Q40: Do these source systems have periodic backups? If so, how often. A: The UNC-GA data warehouse is an Oracle database that uses the combination of cloud control and data guard to handle D/R failover along with nightly backups.

Q41-49: Error in numbering

Q50: What is the total number of users (at System level and campuses and external) that will access the dashboards?

7 of 25

Page 8: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: See Q8.

Q51: What is the Number of Mobile users? A: See Q8. Assume some reasonable fraction (comparable to normal patterns of web browsing) of the public/anonymous users would be on a mobile device.

Q52: What is the Number of concurrent users will access application? A: See Q8.

Q53: What is the total number of users at System level that will have drill down access ? A: See Q8.

Q54: What is the total number of users at each campus that will have drill down access ? A: See Q8.

Q55: What is the total number of power users at System level and at each campus that will have drill down access? A:See Q8.

Q56: What is the total number of users at System level that will have query access? A: See Q8.

Q57: What is the total number of users at each campus that will have query access? A: See Q8.

Q58: Given that UNC has a large investment in SAS software, is this solution meant to replace SAS? A: We do not anticipate that a single monolithic solution will be able to meet our diverse set of needs in this analytics space. We specifically wrote the RFP in search of a “platform” that will allow for an ecosystem of tools to leverage, and we anticipate SAS will be one such tool.

Q59: How many users does UNC-GA expect for the Data Analytics Platform? Will the projected number of users increase over an anticipated timeframe? A: See Q8.

Q60: "State reserves the right to extend a contract term for a period of up to 180 days in 90-day-or-less increments." Software maintenance is typically billed annually in advance. Does the State anticipate this term to apply to software maintenance and/or does the State expect vendors to bill software maintenance other than annual? A: UNC GA does not anticipate the 180 day term applying to software maintenance. UNC GA anticipates extending the term in one year increments.

Q61: Does the State request pricing for software, services, and maintenance for the option years? (attachment C) A: Yes, we want to understand the ongoing, recurring costs.

8 of 25

Page 9: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Followup Question: Would Proposal final pricing include software license and infrastructure cost as well? A: Yes

Q62: The RFP’s Objectives section refers to the creation of Dashboards for admissions, enrollments, financial aid, Student success, budgets and finances in a 4-month window: Is there a specific number of dashboards and or reports that will be in scope for the initial phase of the implementation? Are you able to provide a description of these in scope dashboards; visualization expectations, number charts/graphs per dashboard? A: See Q6.

Q63: Are you open to a cloud (SaaS) Analytics platform solution?

1. Does UNC-GA have a preference to on premises or Cloud solution? 2. Does UNC-GA have a preference for a Cloud provider? 3. What is your cloud strategy for this Analytics project? 4. Are you open to having additional data tables, extending the data warehouse, hosted in a cloud

environment? 5. If the solution is provided in a SaaS methodology would you prefer to have the vendor manage both the

operating system and the software applications? A: We are open to both an on-premise and cloud solution; if a cloud solution we currently do not have a preference for cloud provider, but that determination would be within scope of this award. We expect that some extension of the warehouse would be necessary in a cloud environment and we would be looking for the vendor to manage both the systems and applications layers within the cloud.

Q64: If the proposed solution is on premise, should the vendor include infrastructure cost (server, network connectivity, etc.) and must be supplied by the vendor? A: To propose an on-premise solution, the vendor should identify all necessary infrastructure requirements along with approximate costs of each. We would evaluate the hardware requirements and costs against our existing capacity and with our existing hardware vendors to determine the approach most suitable to supplying the infrastructure.

Q65: In the objectives section (5.2), sections 6 and 7, You refer to the need of expanding the current data warehouse capabilities to other constituent UNC institutions.

1. Is there an estimated database size (in TB) for the extended analytical platform that will support these additional datasets?

2. Do you have an estimated number of additional tables, ETL jobs, or data volumes that will be transferred to the new platform?

A: We would be looking to import around 200GB for the existing data warehouse. We are planning for an increase of about 40% each year, which puts it about 3-5 TB in 10-12 years. It is unknown how much additional storage would be required if campuses join and want to load their own information or if GA wants to start using it to data sources other than the data warehouse. Therefore, require the storage solution to highly elastic (both up and down) so that we can easily grow over time; furthermore, we would expect the costs in the proposal to be in $/TB or some similar metric to allow for this elasticity. Followup Question: Answer says “we would be looking to import approx. 200 GB for existing data–warehouse”. Would these data to be loaded from current Data–warehouse to Analytic Platform on incremental fashion? A: I think the answer depends on the nature of the platform and the technology being used. However, yes, it would be ideal to implement this as an incremental load as the materialized views grow naturally over time.

9 of 25

Page 10: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q66: UNC-GA current data warehouse;

1. What database (software vendor, version) is the current data warehouse? 2. What is the current data warehouse location?

i. On premises – locations? ii. On Cloud – Cloud provider?

1. Are there any considerations to moving your existing data warehouse to a Cloud environment, in order to improve performance of the overall solution or reduce cost?

A: Oracle 12c, on-premise. The current architecture is not scaled to do the high-volume reporting we are expecting with this analytics platform, so we would need to invest in additional hardware or move to the cloud to achieve performance we expect.

Q67: In reference to the statement; “allow for all of the UNC constituent institutions to benefit from a baseline set of services” (5.1 scope of Work – General)

1. How many UNC constituent institutions are there? A: 17. See http://northcarolina.edu/content/our-17-campuses

Q68: Does UNC-GA have a specific target budget in $ USD for the project for either total or for the following individual items?

1. Infrastructure 2. Software 3. Training 4. Implementation services

A: UNC-GA is looking for a proposal that provides the best value. We operate in a highly constrained budget environment but have not set specific dollar targets for this RFP.

Q69: Data Analytics platform; (5.1 scope of Work – General)

1. What type of statutory reporting is in scope for the RFP? 2. What type of data visualizations are required/anticipated? 3. Are there any open score tool preferences for data science/diagnostics?

A: In the near-term all of our statutory reporting is handled via other means; depending on the success of this platform they may be converted over time to use it. We do not want to re-invent the wheel, so a significant part of the RFP is having vendors suggest to us the best practices for visualizations, data science/diagnostics and the corresponding tools needed to deliver them.

Q70: How many users to you anticipate for;

1. statutory reporting 2. dashboards/visualizations 3. science/diagnostics 4. predictive analytics

A: See Q8

Q71: How many users do you anticipate on the project team that should be trained, power users, administrators etc.?

1. statutory reporting 2. dashboards/visualizations 3. science/diagnostics 4. predictive analytics

10 of 25

Page 11: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

5. report developers A: See Q8 and Q27

Q72: Are you able to provide policies as they relate to this project, in reference to: (3.1 Method of Award)

1. Compliance with information technology project management policies 2. Compliance with information technology security standards and policies

A: Primary focus will be on substantial conformity with the specifications and other conditions set forth in the solicitation.

Q73: What are the ETL implementation Services requirements?

1. Will the implementation Services include any additional ETL jobs? if so, please provide details of expectations.

2. Are the ETL jobs going to include data quality algorithms A: It is our assumption that the selected platform will require some form of ETL to get access to our existing data warehouse data. Our existing data warehouse handles data quality already (see Q28) by a series of data validations and corrections at the source (Banner/PeopleSoft).

Q74: What is the current Data Archiving UNC methodology?

1. What data archiving solution is UNC currently using? 2. Is the data archiving requirement going to be applied only to your existing data warehouse environment,

the new (extended) platform or both existing and new (extended) platforms? A: See Q40. The backups referenced are stored to tape. It is a good question about the extended platform and heavily depends on the nature of the selected solution. If the platform contains just optimized copies of the original data, then we would likely not need to archive it; however, if it contains original data that could not be re-constructed from archived sources, then it would likely need to get archived.

Q75: What type of data science/diagnostics are currently in place at UNC-GA?

1. Are there specific tools, open source or others that are required? 2. Are you able to explain your data science/diagnostics requirements?

A: We have limited data science and diagnostics currently occurring at UNC-GA. Policy analyses are primarily supported by SAS programming on specific policy questions/topics, and in a few cases, collaborations between Data & Analytics and other units in UNC-GA have employed STATA or R. Part of the goal of this project is to leverage intuitive BI tools to allow “power users” access to data to do their own ad hoc reporting, thus freeing the time of the programmers/analysts to do more data science and diagnostics.

Q76: Does UNC-GA have any existing predictive modeling tool that’s being replaced?

1. Is there an existing set of predictive algorithms to be satisfied, or future considerations for particular algorithms?

2. Do you have specific Use Cases that predictive modeling would be used for? If so, provide details. A: No existing modeling to replace. Predicting enrollment would be high on the list of needs; other potential areas (not exhaustive list) would include predicting financial aid consumption, retention, course section capacities. We need a tool that will enable us to model not only these things, but the future problem that we haven’t identified yet.

Q77: What is the data architecture (i.e. third normal form or conformed dimensional model) of the current data warehouse(s)? A: The data model is architected in what one would call a conformed dimensional model approach.

11 of 25

Page 12: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q78: Will the vendor have access to the uncleansed version of the source data? A: No Followup Question: in clarification document – Will vendor have access only to current DW (Materialized views and Data Models) as a source to be loaded into new Analytics Platform? A: Yes

Q79: Current refreshes occur 13 times per year – is there a desire for a more frequent data refresh rate and if so, can you please specify the interval? A: For official reporting, no; our existing data warehouse can handle any interval of collections we need and this is the appropriate frequency. However, we do currently pull a nightly snapshot of most datafeeds, and we would be interested in potentially incorporating that into the analytics platform to handle more daily management types of reporting. Note that data in this fashion would be considerably “less clean” than the official reporting but potentially useful for operational decisions. Followup Question: Based on answer, do we assume that daily incremental load and monthly aggregates need to be made available within new Analytic Platform? A: Yes

Q80: What toolsets are currently in use for existing data analysis and reporting? Please include names for reporting, ETL, analytics, visualization and other functions. A: See Q28.

Q81: Will all solution infrastructure needs (storage, computing) be incremental or does UNC GA have existing capacity that can be utilized? A: See Q66.

Q82: What projected lifetime should the vendor base it's total cost of ownership estimate? A: We don’t have a specific time frame in mind. We would expect to have the solution in place at least five years, probably more. Articulate in your proposal why the time frame you suggest makes sense.

Q83: Please provide an estimate of the expected user base broken by GA and campus/constituent institution, itemized by number of analysts, power users and regular users. Please provide the same for 2 year estimated rate of growth. Please specify concurrent vs. total where applicable. A: See Q8.

Q84: Have the required KPI's been defined and will the required underlying data be available immediately? If not, when will this information be provided and available to the vendor? A: This is related to Q6. In general, no, there are no defined KPIs. We are in the midst of the our strategic plan and we expect KPIs to be defined in the process, which will happen in parallel to establishing the analytics platform. The analytics platform needs to be sufficiently flexible to allow us to define (and eventually redefine) KPIs as these will change over time.

12 of 25

Page 13: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q85: What are the regulatory expectations, limitations or constraints with regard to the delivered solution(s) (such as FERPA, HIPAA, etc.)? A: The analytics platform will have data that needs to be FERPA protected. UNC-GA does not currently have data that requires HIPAA protected, but campuses do. It is unclear whether those data would ever be used by the participating campuses, but ideal the analytics solution would eventually allow for this possibility, albeit not a hard requirement.

Q86: Vendor asks UNC-GA to supply a network diagram to include all architecture supporting the current data warehouse referenced, can you comply? Hardware make, model, drive types and specifications for storage (NAS/SAN/DAS), quantity, make & model of servers with specifications including CPU cores and RAM, network topology with switch infrastructure as well as operating systems & hyper visors must be included. A: No, not specifically. See Q28 for a high-level architecture. The vendor can assume any new infrastructure requirements for an on-premise solution would be new. Our existing infrastructure is Dell-based servers, attached to Dell-based SANs, running VMWare with Cisco networking gear.

Q87: What is the current size of the data warehouse, capacity utilized and capacity for growth? A: See Q65.

Q88: What is the intended size of the data warehouse years 1, 3, 5, 7, & 9? A: See Q65.

Q89: How much data do you intend to pass through the platform once deployed? At what throughput and at what speeds? A: See Q65 for the size of data. Our current system runs on a 10G line to a hybrid SAN. At peak times we process 1TB of transactions going in and purging from the system. The high for IOPS is 1881 and our median is 40 for the last 30 days which includes a submission period due date which is heavier in transactions. The avg write to read is 73% write and 27% read.

Q90: Are any other analytical databases in use, and if yes, can network diagrams be provided as described above? A: No

Q91: Can you provide reporting on peak data transfer rates from the data warehouse? A: Outside of the details provided in Q89, more specifics would need to be given as to what reporting on peak data was desired. We do collect statistics from the SAN and database server on a variety of things.

Q92: What is the expected wait time for a particular report/visual/dashboard? A: We expect extremely snappy and responsive interfaces. While we don’t have a specific metric set in stone, we expect a pleasant user experience, which would typically mean reports, visualizations, etc. are rendered in 2 seconds or less.

Q93: Will visuals/dashboards appear in other applications? Iframes, Webparts? Etc. A: We would like to be able to embed any visualizations created with the BI tool to any of our existing web-based websites/portals. Therefore, some kind of web-based API (javascript or equivalent) would be highly desired.

13 of 25

Page 14: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q94: What data transformation activities are expected? A: See Q73.

Q95: What predictive language(s) does UNC expect to use? A: We have no expectations in this area. We want the solution that will best match our needs and existing human capital talents and seek recommendations from the vendor.

Q96: Will UNC-GA accept an RFP response that requires adherence to minimum performance and specification guidelines published by the OEM for the subscription service? For example, UNC-GA owned infrastructure must meet or exceed established OEM performance metrics or upgrade to meet the requirements to allow for 99.9% availability. A: Yes.

Q97: Can you provide a complete description of the current change control process for all infrastructures that support the performance of the data warehouse described in this section? A: Change Control in terms of patching/maintenance is processed via ticketing. Development systems are maintained similar to production systems so if patching does not require a reboot, we flip systems to standby and patch the server / database in question and then flip back. Zero down-time is the goal in all of our patching and maintenance but if not feasible then there is sign-off needed by several customers as to a date, time and length of window that has to be agreed upon and coordinated. The ticket created is used as our communication workflow between DBA and systems groups to communicate completion and handoff to the next person until the ticket is complete.

Q98: In the first paragraph, you mention the goal is for this to be used to expand data science/diagnostics and predictive modeling. Can you expand on this goal? A: See Q75, Q76.

Q99: In the first paragraph, you mention the goal is for this to be used to expand data science/diagnostics and predictive modeling. Can you expand on this goal? A: See Q75, Q76.

Q100: In the first paragraph, you mention the goal is for this to be used to expand data science/diagnostics and predictive modeling. Can you expand on this goal? A: See Q75, Q76.

Q101: What is the purchase intent of this RFP? Will this be contracted at the system level with additional access licenses made available to the institutions? A: Yes.

Q102: How long is the mandatory meeting on the 28th? A: We have reserved 2 hours to allow for sufficient time to answer questions and ensure clarity. However, we do not expect it to last the full duration.

Q103: Based on the answers at the meeting on the 28th, we will have the ability to ask additional questions based on information shared?

14 of 25

Page 15: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: After the meeting on the 28th we’ll accept additional questions until 4pm ET on 11/1/2016. Answers will be posted by 4pm ET on 11/4/2016

Q104: What additional data sources (outside of the student data warehouse) are intended to be included? A: UNC-GA would have a few additional data sources to link to the data mart - e.g., data from K12 and Community College systems used for various state and federal reporting, and data from the National Student Clearinghouse. Constituent institutions may have a wide range of other data to incorporate. See Q26 for examples.

Q105: Is there an interest for delivered predictive models (vendor developed) and available through this solution? A: Yes.

Q106: Do you have any preference for an analytics platform? If so, what are some of the platforms you have considered and why? A: We do not have a preference for a particular platform at this time.

Q107: What were some of the challenges faced, or shortcomings, of using UNC’s current systems’ Visualization/analytics features? A: The needs articulated in the RFP section 5.2 summarize well many of the challenges faced in developing the ideal solution. In particular, items 2-4, 6, 7, and 9 of that section are informed by such challenges.

Q108: Is there a data model in place for the 750 data elements and to what extent are these data elements documented for the purposes of being able to design dashboards with KPIs, i.e., are there existing designs in place for the dashboards such as sketches, wireframes, etc.? A: The data dictionary is available at: https://uncdm.northcarolina.edu/sdm/dictionary.php. The excel version is located here: The data elements, definitions, valid values and validations are documented. Extensive review of definitions occurred via the system-wide data warehouse implementation 2014-2015. There are minimal sketches or high level requirements of dashboards in place. We would look to a vendor to propose some best-practice or standard visualizations.

Q109: What is the expected scope for training UNC staff second to providing materials/documents? A: See Q29.

Q110: Specifically what kind of data are we talking about here and in which form? A: See Q108.

Q111: What is the data volume and for which period? A: See Q65.

Q112: What are the number of dashboards required? A: See Q6.

15 of 25

Page 16: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q113: Does “among institutions” signify only under UNC? A: Yes.

Q114: Depending on user accessibility, the user will be able or unable (i.e. hide) to see columns of data in the front end? Please clarify. A: The original statement in the RFP was ““The product should be able to mask small cells to maintain privacy for those affected populations.” It is our practice that when small cells are displayed in reports (for example less than 5), we do not actually show the number because doing so could allow the report use to determine FERPA-protected information for individual users based on the known demographic makeup of the population.

Q115: What is the significance of “non-proprietary” here? A: As a general principle, we at UNC believe in open standards. We do not want a solution that is exclusively tied to one particular vendor in one particular technology. Instead, we will give preference for solutions that can interoperate with other products using well-defined, open (if possible) standards. However, we expect there to be some element of proprietary software in the ecosystem.

Q116: Which data is being referred to with “beyond the data currently residing…”? A: See Q104.

Q117: According to Section 1 “UNC-GA currently has a data warehouse with deep integrations into the individual campus Enterprise Resource and Planning (ERP) systems, Banner and Peoplesoft.” Since a data warehouse from heterogeneous sources is already established, what is the significance of the requirement of adding local data sets? Shouldn’t it be uploaded in the existing system and we collect data from their data warehouse? Please explain. A: See Q26, Q36 and Q104 Followup Question: Need a better clarity on answer to this question. A: As an example, a campus (or GA) might have a dataset that does not reside in Banner or PeopleSoft, but wishes to be able to upload it, connect it to the “official” ERP data and use it in a data visualization, a dashboard, or a predictive model. UNC-GA does not want to become a middle-man needed to build a custom such area inside our existing data warehouse to do nothing more than push it into the analytics platform. We see the analytics platform to provide this level of data ecosystem to combine the “official” (highly structured, consistent, and cleaned data) data warehouse data with other data sets.

Q118: We would need to understand the existing SAML2 as SSO solution structural hierarchy for proper role assignment from UNC to institution(s) to campus(es) level. A: See Q30.

Q119: Our understanding of Engagement Objective, please validate: The new data analytics platform will leverage UNC’s existing data integrations & data warehouse application that will expand their existing Data warehouse platform into data visualization, data science and predictive modeling platform. This platform should allow all their UNC constituent institutions to benefit from a baseline set of services and for any of the UNC constituent institutions to optionally leverage the platform for its own data analytics needs as well. Is our understanding correct?

16 of 25

Page 17: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: Yes.

Q120: We understand that overall scope of the engagement is to build new Data Analytic Platform leveraging UNC’s existing Data-warehouse architecture. The new Data Analytic Platform should be able to generate pre-defined KPI dashboards and provide ability to perform data analysis to meet business needs. Please clarify. A: This is correct.

Q121: We understand that new Data analytic platform might/will require re-designing and re-populating existing/current Data warehouse platform to meet business requirement needs. This would require database/data-warehouse re-modeling, re-writing ETL Scripts and re-defining data load strategy. Is that correct? A: No. Our intent is to capitalize on the existing data model and extensive data integration work already undertaken.

Q122: We understand that building Predictive/Statistical Models on top of new Data Analytic Platform will be a part of engagement scope and detailed model build requirement will be provided during requirement elicitation phase of the project. Is that correct? Can you provide some examples of predictions KPIs that needs to be built as a part of scope? A: Yes that is correct. There are dozens of indicators that should be included and we would want to iteratively evaluate the predictive capability of each variable or “KPI” and update the models on a continually basis in order to continuously improve our models. Of course the examples would depend on the independent variable. If we take graduation or persistence as an example some “KPIs” might be a variety of demographic variables, GPAs, SAT/ ACT scores, new/ transfer, program enrollment, financial aid received/ used/ lost, types of aid, transfer institution, credit hours attempted, D/F/W status, course taking patterns, etc.

Q123: Apart from ERP systems, Banner and PeopleSoft, please provide the details of other source systems, which are in scope and gets loaded into current data-warehouse. A: The current data warehouse is loading from those ERP systems solely. See Q26 and Q104 for further insights.

Q124: How many years of historical data is available within current Data-warehouse system? What is the current data volume (in terms of GB or number of rows)? What would be the approximate data volume growth every year? A: See Q65 for size/volume information.. The current data warehouse system went live during the 2015-16 academic year, and UNC-GA is loading historical data back to the 2009-10 academic year.

Q125: What is the data load process within current Data-warehouse (Incremental, Snapshot, or Destructive)? Is there a process of change data capture in current ETL process that loads data into data-warehouse? A: The current process is snapshot-based. Unless there is compelling case presented (which we would be open to if it provides significant benefit), we do not intend to change the underlying ETL process that loads the data warehouse.

Q126: How many hierarchies (for different dimensions/masters) are being maintained in current data-warehouse system? Please name them. A: Mostly all of our dimensions are key/value pairs that do not form levels of hierarchies. The only hierarchies that could be formed would be time, degrees, fund_codes, ceeb_codes, test_components. We do not model our dimensions with hierarchies for OLAP cubing. We do model them with primary and surrogate keys to capture history if needed.

Q127: How is the data accuracy of source systems that feeds data within current Data-warehouse? How frequently does the source systems get refreshed/updated? Do we have manual/automated updates process defined for the in-scope source systems?

17 of 25

Page 18: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: See Q171 The main “source systems” (Banner/ PeopleSoft) are for the most part live, transactional systems or at the very least are updated multiple times daily from the live, transactional system. Other data sources have a wide variety of update frequencies. One of the implications of the architecture laid out in Q28 is that data quality issues must be corrected at the source.

Q128: How stable are the underlying source systems i.e. are any changes or design enhancements gets done to these operational database systems on time to time? A: Relatively stable, and yes, individual campuses make changes/updates to their respective ERP systems from time to time.

Q129: What is the size of current (in scope) by source data systems - in terms of data volume? A: See Q65.

Q130: What is the current process of data upload into each of the in-scope source database? Is there any spreadsheet extract that gets uploaded within source systems? A: See Q28, Q127

Q131: Are there any issues with current source systems (DQ, Duplicates, data integrity & conformance, Manual Updates, Refresh frequency, incomplete dataset etc)? Does ETL team spend considerable amount of time in source data profiling/cleansing before loading data within current Data-warehouse? A: See Q127. Yes, there is an extensive data cleansing process that must be passed before it is accepted into the data warehouse’s data model.

Q132: What is the geographical coverage of data that needs to be populated into to-be Analytic Platform? Is it limited to a particular region Ex. Specific region outside USA region? A: In general, North Carolina.

Break in numbering.

Q135: RFP says, “The UNC-GA seeks a data analytics product/platform and associated services to leverage existing data integrations”. What is the approximate ETL (data integration package) count built for existing DWH? A: One per data feed that would include several stages (Pull data, Stage Data, Validate Data, Migrate Data, Materialize Data). Each ETL will handle all records load/reload, and one to many record(s) update, delete, add process for changes.

Break in numbering.

Q137: Are there any joins between multiple source systems, that are written to extract data from existing source systems? A: The joins that exist between multiple sources are not in processes in the extraction from the source system but would need to occur in the data warehouse or in the vendor proposed analytics solution.

Q138: Are there any one time historical data load requirements within current data-warehouse system? If yes, How many years of data will have to be loaded/maintained in Data warehouse repository? A: See Q124

18 of 25

Page 19: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q139: Is there a need of data aggregation process before loading data into the data warehouse? A: Maybe, it depends on the solution that the vendor proposes.

Q140: What is the current load Window for ETL data load process within current data-warehouse repository? Are there any issues with SLA w.r.t. Data load window? Is there a need to optimize data load window for data load process within current data-warehouse repository? A: The current data load is both a scheduled and ad-hoc process. The scheduled load happens between 12:30 am and 7am. The majority of the load processing time depends on the response time of the campus database systems. Data can be refreshed at will on an ad-hoc basis throughout the day for open submissions. The current process would not need additional optimization for the purposes of this RFP. The data movement for purposes of this RFP would be a scheduled process that would most likely need to be pushed as late as possible but before the 12am nightly refresh window.

Q141: FRP states data gets loaded 13 times a year within current data-warehouse repository. What is the current data load frequency (daily/weekly/monthly/quarterly)? Is there a need for any intra-day/near real time data integration requirements with source system? A: See Q145

Q142: How is data rejections handled in current Data warehouse architecture? Is there a notification engine in place to alert/notify the system administrators of the rejects & dependencies in source data (e.g. unavailability of a source file before start of the load cycle)? If no, is there a need to implement similar functionality? A: See Q131. There is already interfaces and alerts to handle all of the data quality issues that are discovered.

Q143: What is the current ETL technology used for data integration process? Does existing data integration technology support scheduling of jobs, failure alerts? If no, is there a need to implement similar functionality? A: See Q28. We currently have job scheduling and alert failures.

Q144: Please outline/validate the current technology landscape (with versions) by providing answers to the below mentioned pointers (relevant to the scope of this exercise)? Database: Oracle, version-?? ETL: Oracle Procedures OLAP/Reporting Tool? A: Oracle 12c, combination of PL/SQL and PHP for ETL, no OLAP. See Q17 for additional details.

Q145: Do we have time dimension defined as a part of current Data warehouse architecture? If yes, what is the lowest member/level in time dimension hierarchy (Day, Week, Month) A: Snapshot run is the time dimension (at least once daily) for a subset of reporting. The main reporting snapshots occur 4 times per term (beginning of term, census, end of term, post-grades).

Q146: How are we handling slowly changing dimensions using ETL process within current data-warehouse architecture (Type 1,2,3)?

19 of 25

Page 20: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: Dimensions will be updated via administrative interface done by UNCGA to handle the slowly changing dimensions. The ETL pertaining to the RFP would pull the dimension table changes just as they would the raw data. We would be using Type 2.

Q147: What is the high-level list of Dimensions & Facts that we have in existing data-warehouse landscape spread across different subject areas. Also, please provide information on hierarchies for each of the dimensions. A: See Q126, Q108.

Q148: How many dimensions would have a need to track the history of change (SCD 2 scenario)? This has a significant bearing on the data model and system complexity. A: It should be assumed that all dimensions have the ability to track history.

Q149: Please provide examples of metadata that is currently being captured in the metadata repository. Metadata can be like Date and time of data load; Number of rejected records; Total number of records processed etc. A: See Q108. We capture every single import and export of data into the warehouse. We capture date/time of load, number of records, errors, warnings, submission notes (expansive error messages), the submission snapshot it belongs to, the type of submission it is.

Q150: Is it possible to provide logical data model and technical landscape (data flow) for existing database/data-warehouse structure? A: See Q28, Q108.

Q151: Is there a need for addition of any new source/transactional systems within existing DW/Data Repository than what it already has now to meet new business needs? A: Multiple data sources from within and outside of the existing data warehouse will need to be integrated. UNC will need the flexibility to add additional data sources over time.

Q152: Which reporting tool is being used/purchased as a part of the commercial license for UNC-GA? A: See Q17.

Q153: RFP states, “Selected product should be compliant with a SAML2 as a Single Sign On (SSO) solution.” Is there a need to integrate reporting portal with in-house portal application for displaying, scheduling and accessing the reports or new reporting tool portal would be used as platform to view reports/dashboards? A: See Q30.

Q154: Will these be a need to send reports to users email inbox or is the user expected only to login to the portal/ report server and create/view/edit the report? A: See Q159

Q155: Does a reporting template or expected format exist or it needs to be formulated during requirement definition? A: There is no expected format. See Q159

20 of 25

Page 21: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Q156: What is the count of existing set of reports? Will it be possible to re-design existing set of reports to meet new business BI needs? A: Broadly defining reports as the range of routine outputs used for internal data validation, state and federal reporting requirements, and various other recurring monitoring and reporting efforts - and multiplying many by 16 or 17 institutions - there are hundreds of reports supported by the data warehouse. Yes, it is possible to re-design existing reports, and desirable to do so in many instances.

Q157: Will it be possible to share critical existing reports during requirement definition phase of the project in order to understand reporting requirements & needs? A: Yes. However, the purpose of this undertaking is to go beyond UNC’s current reporting and capabilities and not to recreate what we already have in a different format.

Q158: Is there a pre-defined performance/SLA standard expected for Report Refresh time? If yes, would it be possible to provide a high level SLA for report refresh time? A: See Q164

Q159: What are the output formats that is expected for reports for end users to view? Ex. Pdf , .excel, e-mail etc. A: Ideally, we would expect an analytical and visualization solution to be able to present data in both static and interactive forms and to allow for user download in multiple formats, including Excel, PDF, text files, image files etc.

Q160: What are different data securities needed for the engagement? Is there a requirement of row level data security in Reports? For example, Can all users view all reports contents or rows need to be filtered out based on users privilege? Does existing DW/BI architecture include access control and security for end users? A: Row level (filtered) security would be needed. Some users could under FERPA have access to student data for a particular population but not for every student. Additionally for data visualizations and analytics, some individuals would need to be able to drill down into unit record data while others would be limited to aggregate data not including small cell sizes (cells of 5 or below must be excluded from view). The existing data warehouse does have various levels of access control but additional security levels not currently included in our security model would likely need to be added for the solution.

Q161: Is it possible to share some sample reports to understand the standard report layout that needs to be used to build new reports? A: See Q157.

Q162: Approximately, how many Analytic and operational reports/dashboards are in the scope of engagement? Approximately, how many KPIs (Key Performance Indicators) would be in scope of engagement? Would it be possible to provide the count of reports/dashboards/KPIs that needs to be built for end users? A: see Q6.

Q163: What is the total count of users who will be looking at new reports/dashboards? Will these users located in a single geography (USA) or will be spread across different geographical locations/time zones? A: See Q8.

Q164: Is there a need for real time/near real time reports as a part of this engagement?

21 of 25

Page 22: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: Daily refreshes of data at a minimum would be expected.

Q165: RFP states “UNC-GA seeks material results in the form of industry leading visualizations and dashboards within four (4) months of the Effective Date of the awarded contract”. Does this mean; UNC-GA needs to start seeing first set of visualization deliverables within four months of awarded contract? A: Yes.

Q166: Is there a need to choose dedicated licensed product for predictive modeling business needs (R, SAS, SPSS, Statistica etc.) or would prediction need to be a part of BI/OLAP BI tool. In case of advanced prediction business needs, it is always recommended to choose dedicated tool stack for predictive modeling. Few BI tools do provide Predictive modeling capabilities but they are very rudimentary in nature. A: UNC is open to a range of approaches, including the above recommendations.

Q167: Are there any data dictionary and data documentation (requirement & design document) in place for current data in data warehouse? Is it possible to share some of the artifacts during the requirement phase of execution in order to understand technical details of current architecture? A: The data dictionary is available at: https://uncdm.northcarolina.edu/sdm/dictionary.php See Q108.

Q168: Where do we capture rejected data and how do we manage them for reload in current environment? Would same process need to be replicated for new Analytic Platform? A: See Q142. At this point, we don’t see it as a requirement to be able to do to this kind of data quality checking in the tool. However, it is possible that this feature could be used in the future for needs that are currently not known.

Q169: What are the key reasons/concerns/business issues/pain points that have triggered the need for this exercise? A: UNC has developed a data warehouse that supports statutory reporting needs as well as some limited data visualization and analytics. The key trigger for this exercise is the desire to leverage our robust data warehouse to increase meaningful and actionable data visualization and analytics.

Q170: What are the major concerns with respect to your current data information & data management? A: See Q169

Q171: What are typical Data-Quality (DQ) issues currently being faced in the underlying source systems that UNC need to rectify? A: UNC has a robust ETL process including hundreds of data validations before data is brought in from the underlying source systems (Banner/ PeopleSoft) into our data warehouse. Therefore, data quality issues are limited from Spring 2016 forward for data in the data warehouse. UNC would like to leverage its data warehouse in this solution. That is of course not to say that data quality issues do not exist. However, they are minimized to the extent that they can be by UNC’s validation process, which UNC is constantly refining.

Q172: Is it possible to share critical success criteria that will be used to evaluate the success of the project? A: Specifics of deliverables, timelines, and measures of success will be easier to articulate after review of proposals and selection of the proposal that provides the set of services detailed in section 5.2 of the RFP (and elsewhere within the RFP and subsequent clarifications/communications, like this Q&A document) at a quality and price that constitute the best

22 of 25

Page 23: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

value overall. GA staff will work with the selected vendor to outline specifics of deliverables, timeline, and success criteria prior to embarking on the implementation. At a high level, success criteria likely would include: 1. Responsive, proactive, and direct communication; 2. Deliverables complete and in production on time; 3. Implementation of all software, hardware, and creation of at least one high value, high-quality dashboard in the first four months; 4. Iterative, continuous improvement and expansion of data visualizations and predictive analytics on a regular and timely basis; and 5. High quality and engaging training.

Q173: RFP states: “The Contract shall have an initial term of one (1) year, beginning on the date of contract award.” What-if Analytic platform gets deployed in production well within 1-year timeframe? In other words, will the project execution be allocated to Vendors based on Annual contract (Implementation + Deployment + Warranty + Support) or will it be based on solution completion (deployment + Warranty)? A: UNC GA anticipates the initial term will have the implementation and deployment costs along with other costs(i.e warranty and support) depending on when the platform gets deployed into production. Costs associated with annual extensions would be more directed towards support, warranty, licenses, etc.

Q174: Should total Proposal Price include any new software licensing cost as well (if required)? In other words, Pricing for Technology Stack Licensing to be included in final price in the proposal? A: Yes.

Q175: Is UNC open to a model wherein:

• Supplier acts as a System Integrator bringing in best of breed products for the data analytic platform • Supplier will be responsible for end to end implementation, integration and maintenance of platform • UNC will purchase the License directly from the respective product supplier(s) The Product supplier will enter into a License agreement with the University

A: Yes. UNC is open to a range of approaches, including the above.

Q176: Please confirm the requirements: c) Completed and signed version of EXECUTION PAGES: We understand that this will include completion of any content requirements and signature in all sections from Section 1.0 Purpose and Background till Section 6.0 Contract Administration – this includes completion of Section 4 and Section 5. However as per the proposal contents list responses to Section 4 and 5 are repeated again j) Response to Requirements. Section 4 and k) Response to Scope: Section. Please clarify if the information is to be repeated. A: Please see question 178 and the corresponding answer.

Q177: Are there any other data sources that need to be accessed besides Oracle that is mentioned in Paragraph 2? A: See Q4, Q26, and Q104 for insights.

Q178: When vendors include RFP pages 2-20 in our proposals, are vendors to respond to Sections 4 and 5 within pages 2-20 or are vendors to provide responses to Sections 4 and 5 in their own templates? A: Vendors can respond to section 4 and 5(j & k in the list of Proposal Contents) within the body(pages 2-20). It is also acceptable to respond to section 4 & 5 in their own template and send the body separately. Refer to section 2.7 for the complete list of required proposal contents.

Q179: Please provide a Word version of the Execution Pages and Attachments D, E, and F.

23 of 25

Page 24: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

A: Please complete and sign the versions that were provided in the RFP.

Q180: Is the 4 month deadline for implementation of entire solution, or is this a Phased Approach? A: See Q8, Q12.

Q181: Please describe "associated" services. Do you need Professional Services to implement the program and build dashboards etc.? A: Yes.

Q182: If the University System of North Carolina chooses to implement via a Cloud provider, will the Vendor be required to supply the Cloud Service or does the University already have a provider of such services? A: See Q.175

Q183: What is the operating system of choice for this implementation? Windows? Linux? Other? A: See Q5.

Q184: Is UNC-GA looking for a COTS solution which already includes the “industry-leading set of dashboards, etc.”? Or is UNC-GA looking for custom development of the solution? If custom development, is UNC-GA looking for the entire solution to be implemented within 4 months, or just a subset of the solution to be completed and rolled out within 4 months? A: We would certainly need the ability to modify and customize any dashboards. The base could the a COTS (consumer off the shelf) including “industry-leading set of dashboards, etc.” That is up to the vendor to propose how our needs can best be met. In 4 months we would like the complete set-up of the baseline product(s), integration of our data into the solution, and a subset of dashboards in production. See Q8, Q6, Q12, Q34 for additional information.

Break in numbering.

Q186: Please describe "campus users." Are they employees of the University System of NC or are they students or both? Do you need a student facing dashboard? A: “Campus users” are faculty and staff that work at one of our 17 constituent institutions that make up the University of North Carolina System. Some faculty and staff may also happen to be students by virtue of taking classes at one of our universities. We might eventually have use cases for students as users or a student-facing dashboard.

Q187: Please elaborate on what "non-proprietary" means to UNC-GA. A: See Q115.

Q188: Would onsite training be the preference for "hands on training”? A: Yes.

Q189: Information Builders (“IBI”) takes exception to Attachment B. Notwithstanding anything to the contrary contained in the RFP or Attachment B, in the event IBI is selected as the vendor of choice, the services proposed by IBI shall be governed by the Master Services Agreement (“MSA”) dated June 30th, 2014 between IBI and the University of North Carolina General Administration and Affiliates, a copy of which will be provided upon request.

24 of 25

Page 25: Analytics RFP - Ve ndor Q uest i ons - BidNet the platform for its own data analytics needs as well.” The RFP is unclear as to ... the response confine itself solely to ... grade

Notwithstanding anything to the contrary contained in the RFP or Attachment B, in the event IBI is selected as the vendor of choice, the Software and Maintenance Services proposed by IBI shall be governed by the Master Software License Agreement (“MSLA”) dated March 19th, 2002 between IBI and the University of North Carolina General Administration and Affiliates, a copy of which will be provided upon request. IBI agrees that it will enter into good faith negotiations with the University to address any required revisions to this agreement to include any additional mutually agreed to terms and conditions that address specific requirements for this RFP. A: UNC GA rejects and will not be required to to evaluate or consider any additional or modified terms and conditions submitted with vendor's proposal. Refer to Section 2.2 in the RFP

Q190: Please describe in clarity expectations of "Hardware Support." A: If the vendor’s proposed solution is cloud-based, we would expect all hardware support to be performed by the vendor. If it is on-premise then we would expect that level of support to occur with local UNC-GA resources (unless there is a highly specialized piece of hardware for which we are unequipped to support).

Q191: Will UNC GA provide their own hardware? A: It depends on the proposed solution, but it is a possibility for UNC-GA to provide our own hardware. That is an option, but we would need to understand the hardware and maintenance requirements to assess cost, staffing needs, and overall feasibility. See Q.175

Q192: IBI takes exception to these Supplemental Terms and Conditions for software as a service (SaaS). In the event IBI proposes hosting, the parties shall enter into good faith negotiations to address terms and conditions specific to hosting services. A: UNC GA rejects and will not be required to to evaluate or consider any additional or modified terms and conditions submitted with vendor's proposal. Refer to Section 2.2 in the RFP.

Q193: Are Customer Success Stories allowed as part of our response to the RFP? A: Yes.

25 of 25