the power of analytics in cost management - pdi 2016 · 2016-06-30 · analytics in cost management...
TRANSCRIPT
THE POWER OF ANALYTICS
IN COST MANAGEMENTNational PDI 2016
03 June 2016
Mr. David Molinari, HQDA
Ms. Denise Oberndorf, CALIBRE
WHAT IS COST MANAGEMENT?
Purpose of Cost Management
Provide the ability to make resource informed decisions
Resource informed decisions:
• Allow organizations to identify
more optimal solutions,
• Provide an ability for proactive
management, and• Inform trade off decisions and
impacts.
WHAT IS ANALYTICS?
Analytics
The discovery, interpretation and
communication of meaningful patterns
in data
Relies on the application of statistics,
computer programming and
operations research
Includes data visualization
ANALYTICS IN COST MANAGEMENT
Understanding Value
Budget and spending information is not sufficient
Must be able to understand impact to benefits with
each change in cost (value)
Understanding Costs
Cost Driver: Unit of activity that causes change in
cost
Methods for identifying cost driver include scatter
plots and regression analysis
Understanding Performance
Performance Measurement
Identify key performance indicators and goals
Measure progress toward a desired outcome
DATA STRUCTURES MATTER
Data structures
Format for organizing and storing data
Typically specialized
Impacts transparency and efficiency of data access
Ideal Characteristics
Dynamic: Allows data to inform multiple decisions Rubik's
Integrated: Allows for data to be combined
Efficient: Allows for information to be easily accessed without overburdening the user
DATA MODELS AND APPROACH
Model Types
Forecasting Models
Process to forecast future based on past and present data
Forecast something we can’t control
Time-series methods are most common
Moving Average
Autoregressive
Trend Estimation
Examples of data for forecasting
Price of Electricity
Demand of a good or service
Structural Models
Process to develop a mathematical model to describe how two or more observable variables relate to one another
Influence and inform decisions
Cross Sectional method is common
Examples of observable variables to compare
Consumption of electricity and cost of electricity
Operational hours of a training facility and readiness
FORECAST EXAMPLE: ELECTRICITY PRICE FORECAST (1 OF 6)
First let’s collect historical price data
FORECAST EXAMPLE: ELECTRICITY PRICE FORECAST (2 OF 6)
Try Trend Analysis: $month = A + (Δ$/month) x (Month)
FORECAST EXAMPLE: ELECTRICITY PRICE FORECAST (3 OF 6)
Next let’s try autoregression (AR): $Month = A + Anchor x ($Month-1)
FORECAST EXAMPLE: ELECTRICITY PRICE FORECAST (4 OF 6)
The results of autoregression? Don’t forget out-of-sample testing.
!?
FORECAST EXAMPLE: ELECTRICITY PRICE FORECAST (5 OF 6)
AR with Trend: $Month = A + Anchor x ($Month-1) + (Δ$/month) x (Month)
FORECAST EXAMPLE: ELECTRICITY PRICE FORECAST (6 OF 6)
Comparing forecasts
STRUCTURAL EXAMPLE CASE: BASE MANAGEMENT (1 OF 2)
Drivers:
Energy Consumption – How much do we consume?
Readiness – How many hours can operate training support facilities?
Life, Health, Safety – How fast can we respond to emergencies?
Morale & Welfare – How much does it cost per person?
Trade-offs we face:
What are the priorities? Where can we take risk?
Are we willing to trade training for safety? ….for morale?
How much do we need to budget? What are the impacts if we don’t?
Can we identify best practices and spot potential problems?
STRUCTURAL EXAMPLE CASE: BASE MANAGEMENT (2 OF 2)
Integrating risk allows us to use
structural models to tell us:
How confident we can be
that our budget is sufficient to
achieve desired outcomes
What outcomes we can
expect given our budget
Simulation tools are used to
combine error distributions of
different shapes
RISK!
VISUALIZATION OF ANALYSIS
Visualization
Present data in a pictorial or graphical format
Visualize difficult concepts
Identify new patters
Decision Support
Improves response times
Provides information in a simple format
Increases transparency
Enhance collaboration
VISUALIZATION EXAMPLE Study results on optimizing
response times in Ann Arbor, Michigan
Reduce fire house locations from 5 to 3 and improve response times
Red dots are all fire locations over the last decade
Shaded area represents response times
Station 3, 6 and 4 will close (grey hats)
Station 2 not currently being used, will reopen
Source: The Ann Arbor Chronical
VISUALIZATION EXAMPLE
World coal consumption by region
Automation can help display changes over time
Source: U.S. Energy Information Administration, International Energy Statistics
Link: Rising Asian demand drives global coal consumption growth - Today in Energy - U.S. Energy Information Administration (EIA)
CONCLUSION
Power of analytics in cost management
Better decisions, faster
Predict future budgets and impacts
Informed trade off decisions
Visualize information for easy
digestion