c04 l08 feature selection - amazon web services

11
Feature Selection Brock Tubre INSTRUCTOR

Upload: others

Post on 25-Oct-2021

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: C04 L08 Feature Selection - Amazon Web Services

Feature Selection

Brock TubreINSTRUCTOR

Page 2: C04 L08 Feature Selection - Amazon Web Services

Feature SelectionFEATURE SELECTION

Feature SelectionSelecting the most relevant features from your data to prevent over-complicating the analysis, resolving potential inaccuracies, and removes irrelevant features or repeated information.

Deciding what to keep and what to get rid of.

Page 3: C04 L08 Feature Selection - Amazon Web Services

ID age fur_color born_month born_month_num breed trick_learned

1 2 brown July 7 terrier 0

2 4 white August 8 shepard 1

3 3 golden June 6 terrier 1

4 7 brown September 8 collie 1

5 2 black June 6 retriever 0

6 6 black May 5 collie 1

Feature Selection ExampleFEATURE SELECTION

ProblemCan old dogs learn new tricks?

Page 4: C04 L08 Feature Selection - Amazon Web Services

ID age fur_color born_month born_month_num breed trick_learned

1 2 brown July 7 terrier 0

2 4 white August 8 shepard 1

3 3 golden June 6 terrier 1

4 7 brown September 8 collie 1

5 2 black June 6 retriever 0

6 6 black May 5 collie 1

Feature Selection ExampleFEATURE SELECTION

ProblemCan old dogs learn new tricks?

Page 5: C04 L08 Feature Selection - Amazon Web Services

ID age fur_color born_month born_month_num breed trick_learned

1 2 brown July 7 terrier 0

2 4 white August 8 shepard 1

3 3 golden June 6 terrier 1

4 7 brown September 8 collie 1

5 2 black June 6 retriever 0

6 6 black May 5 collie 1

Feature Selection ExampleFEATURE SELECTION

ProblemCan old dogs learn new tricks?

Page 6: C04 L08 Feature Selection - Amazon Web Services

Feature Selection ExampleFEATURE SELECTION

ProblemCan old dogs learn new tricks?

ID age fur_color born_month born_month_num breed trick_learned

1 2 brown July 7 terrier 0

2 4 white August 8 shepard 1

3 3 golden June 6 terrier 1

4 7 brown September 8 collie 1

5 2 black June 6 retriever 0

6 6 black May 5 collie 1

Page 7: C04 L08 Feature Selection - Amazon Web Services

Feature Selection ExampleFEATURE SELECTION

ProblemCan old dogs learn new tricks?

ID age fur_color born_month born_month_num breed trick_learned

1 2 brown July 7 terrier 0

2 4 white August 8 shepard 1

3 3 golden June 6 terrier 1

4 7 brown September 8 collie 1

5 2 black June 6 retriever 0

6 6 black May 5 collie 1

Page 8: C04 L08 Feature Selection - Amazon Web Services

Feature Selection ExampleFEATURE SELECTION

ProblemCan old dogs learn new tricks?

ID age fur_color born_month born_month_num breed trick_learned

1 2 brown July 7 terrier 0

2 4 white August 8 shepard 1

3 3 golden June 6 terrier 1

4 7 brown September 8 collie 1

5 2 black June 6 retriever 0

6 6 black May 5 collie 1

Page 9: C04 L08 Feature Selection - Amazon Web Services

Feature selection is an intuitive step humans take to reduce the number

of features.

Feature SelectionFEATURE SELECTION

Page 10: C04 L08 Feature Selection - Amazon Web Services

Principle Component Analysis (PCA)An unsupervised learning algorithm that reduces the number of features while still retaining as much information as possible.

Reduces the total number of features in a dataset.

Principle Component Analysis (PCA)FEATURE SELECTION

Page 11: C04 L08 Feature Selection - Amazon Web Services

Feature Selection Use CaseFEATURE SELECTION

Problem Technique Why

Data is too large due to the large number of features Principle Component Analysis (PCA) Algorithm that reduces total

number of features

Useless features that do not help solve ML problem Feature Selection Remove features that do not

help solve the problem