SlideShare uma empresa Scribd logo
1 de 25
FEATURE
ENGINEERING
David Epstein
O P E N
D A T A
S C I E N C E
C O N F E R E N C E_
BOSTON 2015
@opendatasci
FEATURE
ENGINEERING
David Epstein
Senior Data Scientist, Socure
Open Data Science Conference, Boston, MA
May 30-31, 2015
#ODSC, @opendatasci
Talk Outline
• What is feature engineering?
• Limits on number of features
• How to select a “good” set of features
• Standard FE techniques
• TL;DR: As we get better and better models,
focus shifts to what we put into them
• FE interacts with other key areas of DS
Feature Engineering
• (My) Definition: Transforming data to create
model inputs.
Raw
Data
Data
Cleaning
Feature
Engineering
Model
Building
Pre-Processing
Data Workflow
Examples from Kaggle Competitions
Netflix
Titanic
Portuguese Taxis
How does it work?
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.29879 0.48994 0.61 0.559
x -0.00923 0.10256 -0.09 0.930
x2 0.98803 0.03672 26.91 3.92e-09 ***
---
Residual standard error: 1.076 on 8 degrees of freedom
Multiple R-squared: 0.9891, Adjusted R-squared: 0.9863
F-statistic: 362 on 2 and 8 DF, p-value: 1.427e-08
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 10.17912 2.92472 3.48 0.00693 **
x -0.00923 0.92488 -0.01 0.99225
---
Residual standard error: 9.7 on 9 degrees of freedom
Multiple R-squared: 1.107e-05, Adjusted R-squared: -0.11
F-statistic: 9.96e-05 on 1 and 9 DF, p-value: 0.9923
Features are engineered everywhere
Things to be explained/measured/predicted
Finance:
EBITDA
Baseball:
Batting Avg.
Politics:
Partisan Performance
The Big Questions
• Seen in this light, FE is ubiquitous (as all
truly important concepts are)
• Any time you construct an intermediate
variable, you’re doing FE
• Two questions naturally arise:
1. How do you construct “good” features?
2. What are the limits on this process?
• I’ll answer the second one first, because it’s
easier….
Limits on Feature Engineering
• In medical studies, social science, financial analysis, etc.,
two main problems emerge
• Eating up degrees of freedom: relatively small data sets
• # of respondents in survey
• # of patients in trial
• # of elections to Congress
• If your data lives in an NxK matrix, you want to make sure that K is
small relative to N
• Relevance to hypothesis testing, emphasis on explanation
• You generally start with an equation defining the relationship
between the key independent and dependent variables
• Other variables enter your model as controls, not really interested
in their functional form
Limits on Feature Engineering
• In most modern data science applications, neither is an
issue
• We start with lots of data, and
• Care more about prediction than explanation
• So why not add in lots of extra variables?
• Think of your data not as what goes into your model, but a starting
point for the creation of new variables, which can then be combined…
Limits on Feature Engineering
• First, adding many correlated predictors can decrease model
performance
• Adding an x4 term to above example actually reduces model fit:
• More variables make models less interpretable
• Models have to be generalizable to other data
• Too much feature engineering can lead to overfitting
• Close connection between feature engineering and cross-validation
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.648123 0.628178 1.032 0.336513
x -0.009230 0.103740 -0.089 0.931593
x2 0.866738 0.139087 6.232 0.000432 ***
x4 0.004852 0.005361 0.905 0.395572
---
Residual standard error: 1.088 on 7 degrees of freedom
Multiple R-squared: 0.9902, Adjusted R-squared: 0.986
F-statistic: 236.1 on 3 and 7 DF, p-value: 2.15e-07
How To Select a “Good” Set of Features
• This is the open-ended question in the field
• Separate (but related to) the question of feature selection
– which variables to retain in a model.
• You can use some metrics to tell which features are
useful, one at a time, like Pearson correlation coefficient
• But this can’t tell you which set of features works best together
• This is an NP complete problem, clearly too computationally hard
• Many new data analysis services include automated
feature engineering as part of their packages
• But if there are features you want in your model, it’s best to add
them in explicitly, rather than depend on these generators.
A “Middle Theory” of FE
• Start with a reasonable-sized set of features
• Include features suggested by domain knowledge
• Test these out individually, build from the bottom up
• Number and type of features depend on model used
• Can include more features if models does some feature selection
• Lasso regression, e.g., logit with ||L1|| regularization (but not ||L2|| ridge)
• GBM with backward pruning (but not random forests)
• Stepwise regression, with either forward or backward feature selection
• Some models are invariant to monotonic variable transformations
• Tree-based approaches divide variables into two groups at each branch
• So, no perfect answer. But there are some standard
techniques every data scientist should have in their bag of
tricks.
Non-numeric to numeric
1. Count # of times each value appears
Zip Code Count
10024 4
63105 2
94304 1
06443 3
10024 4
63105 2
06443 3
10024 4
10024 4
06443 3
Non-numeric to numeric
2. One-hot encoding
Religion Catholic Protestant Jewish Muslim
Catholic 1 0 0 0
Muslim 0 0 0 1
Jewish 0 0 1 0
Protestant 0 1 0 0
Catholic 1 0 0 0
Catholic 1 0 0 0
Jewish 0 0 1 0
Protestant 0 1 0 0
Muslim 0 0 0 1
Protestant 0 1 0 0
Non-numeric to numeric
3. The “hash trick”
string h(string)
“The” 36
“quick” 8
“brown” 92
“fox” 14
“jumps” 75
“over” 25
“the” 36
“lazy” 44
“dog” 21
Non-numeric to numeric
4. Leave-one-out encoding
Single Variable Transformations
x
x2
Log(x)
scaling
x
Two-variable combinations
1. Add: Sum similar-scaled variables
Q1 Q2 Q3 Q4 Total
33 88 51 81 251
11 11 72 30 124
15 36 70 55 176
70 82 8 50 209
99 56 35 86 276
7 20 10 71 107
65 0 25 74 164
96 25 2 89 211
60 29 56 92 238
63 50 96 61 269
Two-variable combinations
2. Subtract: Difference relative to baseline
ViewerID MovieID Date Rating
Days Since
First Rating
44972004 8825 1/1/13 5 0
44972004 0471 2/1/13 4 31
44972004 3816 3/1/13 5 59
44972004 8243 4/1/13 3 90
44972004 2855 5/1/13 5 120
44972004 9923 6/1/13 2 151
44972004 1023 7/1/13 4 181
44972004 8306 8/1/13 3 212
44972004 2771 9/1/13 2 243
44972004 5281 10/1/13 2 273
Two-variable combinations
3. Multiply: Interactive effects
ViewerID Country Domestic DSFR Dom*DSFR
8825 US 1 38 38
0471 CA 0 277 0
3816 FR 0 187 0
8243 US 1 33 33
2855 US 1 87 87
9923 GB 0 42 0
1023 IT 0 192 0
8306 CA 0 365 0
2771 US 1 505 505
5281 FI 0 49 0
Two-variable combinations
4. Divide: Scaling/Normalizing
Country GDP Population
GDP/Capit
a
US 159849 275 581
CA 731812 111 6593
FR 826320 90 9181
IT 573494 80 7169
IR 609223 22 27692
GB 717673 60 11961
NE 605257 15 40350
MX 687944 124 5548
RU 203319 402 506
FI 744983 40 18625
Multivariate/Model-based Methods
1. PCA/Factor Analysis/Clustering: Dimension reduction
Multivariate/Model-based Methods
2. Model Stacking: Outputs of one model are inputs
to the next
• Do this, e.g., on half the data and use as input to the other
half, and vice-versa
(Figure from Owen Zhang)
Conclusion
• Data science grew over the last decade due to
improved modeling algorithms, better cross-
validation procedures
• Now we have a number of good models, can get
at problem from lots of different angles
• Most improvement will come from thinking
carefully about what we put into our models =
Feature Engineering
• This can be (semi-) automated, but it’s still one of
the true arts of the profession
• Domain knowledge remains very important in practice

Mais conteúdo relacionado

Mais procurados

Winning data science competitions, presented by Owen Zhang
Winning data science competitions, presented by Owen ZhangWinning data science competitions, presented by Owen Zhang
Winning data science competitions, presented by Owen ZhangVivian S. Zhang
 
Challenges in Large Scale Machine Learning
Challenges in Large Scale  Machine LearningChallenges in Large Scale  Machine Learning
Challenges in Large Scale Machine LearningSudarsun Santhiappan
 
Feature Engineering for ML - Dmitry Larko, H2O.ai
Feature Engineering for ML - Dmitry Larko, H2O.aiFeature Engineering for ML - Dmitry Larko, H2O.ai
Feature Engineering for ML - Dmitry Larko, H2O.aiSri Ambati
 
Latent Dirichlet Allocation
Latent Dirichlet AllocationLatent Dirichlet Allocation
Latent Dirichlet AllocationSangwoo Mo
 
Deep Dive into Hyperparameter Tuning
Deep Dive into Hyperparameter TuningDeep Dive into Hyperparameter Tuning
Deep Dive into Hyperparameter TuningShubhmay Potdar
 
Ml8 boosting and-stacking
Ml8 boosting and-stackingMl8 boosting and-stacking
Ml8 boosting and-stackingankit_ppt
 
Machine learning with scikitlearn
Machine learning with scikitlearnMachine learning with scikitlearn
Machine learning with scikitlearnPratap Dangeti
 
Feature Engineering in Machine Learning
Feature Engineering in Machine LearningFeature Engineering in Machine Learning
Feature Engineering in Machine LearningKnoldus Inc.
 
Build an efficient Machine Learning model with LightGBM
Build an efficient Machine Learning model with LightGBMBuild an efficient Machine Learning model with LightGBM
Build an efficient Machine Learning model with LightGBMPoo Kuan Hoong
 
Explainable AI in Industry (KDD 2019 Tutorial)
Explainable AI in Industry (KDD 2019 Tutorial)Explainable AI in Industry (KDD 2019 Tutorial)
Explainable AI in Industry (KDD 2019 Tutorial)Krishnaram Kenthapadi
 
Performance Metrics for Machine Learning Algorithms
Performance Metrics for Machine Learning AlgorithmsPerformance Metrics for Machine Learning Algorithms
Performance Metrics for Machine Learning AlgorithmsKush Kulshrestha
 
Dimensionality Reduction
Dimensionality ReductionDimensionality Reduction
Dimensionality ReductionKnoldus Inc.
 
Introduction to natural language processing (NLP)
Introduction to natural language processing (NLP)Introduction to natural language processing (NLP)
Introduction to natural language processing (NLP)Alia Hamwi
 
Introduction to XGBoost
Introduction to XGBoostIntroduction to XGBoost
Introduction to XGBoostJoonyoung Yi
 
Text classification presentation
Text classification presentationText classification presentation
Text classification presentationMarijn van Zelst
 
Machine Learning Pipelines
Machine Learning PipelinesMachine Learning Pipelines
Machine Learning Pipelinesjeykottalam
 
07 dimensionality reduction
07 dimensionality reduction07 dimensionality reduction
07 dimensionality reductionMarco Quartulli
 

Mais procurados (20)

Winning data science competitions, presented by Owen Zhang
Winning data science competitions, presented by Owen ZhangWinning data science competitions, presented by Owen Zhang
Winning data science competitions, presented by Owen Zhang
 
Challenges in Large Scale Machine Learning
Challenges in Large Scale  Machine LearningChallenges in Large Scale  Machine Learning
Challenges in Large Scale Machine Learning
 
Feature Engineering for ML - Dmitry Larko, H2O.ai
Feature Engineering for ML - Dmitry Larko, H2O.aiFeature Engineering for ML - Dmitry Larko, H2O.ai
Feature Engineering for ML - Dmitry Larko, H2O.ai
 
Latent Dirichlet Allocation
Latent Dirichlet AllocationLatent Dirichlet Allocation
Latent Dirichlet Allocation
 
Deep Dive into Hyperparameter Tuning
Deep Dive into Hyperparameter TuningDeep Dive into Hyperparameter Tuning
Deep Dive into Hyperparameter Tuning
 
Ml8 boosting and-stacking
Ml8 boosting and-stackingMl8 boosting and-stacking
Ml8 boosting and-stacking
 
Machine learning with scikitlearn
Machine learning with scikitlearnMachine learning with scikitlearn
Machine learning with scikitlearn
 
Feature Engineering in Machine Learning
Feature Engineering in Machine LearningFeature Engineering in Machine Learning
Feature Engineering in Machine Learning
 
Build an efficient Machine Learning model with LightGBM
Build an efficient Machine Learning model with LightGBMBuild an efficient Machine Learning model with LightGBM
Build an efficient Machine Learning model with LightGBM
 
Explainable AI in Industry (KDD 2019 Tutorial)
Explainable AI in Industry (KDD 2019 Tutorial)Explainable AI in Industry (KDD 2019 Tutorial)
Explainable AI in Industry (KDD 2019 Tutorial)
 
K - Nearest neighbor ( KNN )
K - Nearest neighbor  ( KNN )K - Nearest neighbor  ( KNN )
K - Nearest neighbor ( KNN )
 
Performance Metrics for Machine Learning Algorithms
Performance Metrics for Machine Learning AlgorithmsPerformance Metrics for Machine Learning Algorithms
Performance Metrics for Machine Learning Algorithms
 
Dimensionality Reduction
Dimensionality ReductionDimensionality Reduction
Dimensionality Reduction
 
Xgboost
XgboostXgboost
Xgboost
 
Introduction to natural language processing (NLP)
Introduction to natural language processing (NLP)Introduction to natural language processing (NLP)
Introduction to natural language processing (NLP)
 
Introduction to XGBoost
Introduction to XGBoostIntroduction to XGBoost
Introduction to XGBoost
 
Text classification presentation
Text classification presentationText classification presentation
Text classification presentation
 
Machine Learning Pipelines
Machine Learning PipelinesMachine Learning Pipelines
Machine Learning Pipelines
 
Dbscan algorithom
Dbscan algorithomDbscan algorithom
Dbscan algorithom
 
07 dimensionality reduction
07 dimensionality reduction07 dimensionality reduction
07 dimensionality reduction
 

Semelhante a Feature Engineering

SPC Training by D&H Engineers
SPC Training by D&H EngineersSPC Training by D&H Engineers
SPC Training by D&H EngineersD&H Engineers
 
Data_Preparation.pptx
Data_Preparation.pptxData_Preparation.pptx
Data_Preparation.pptxImXaib
 
Influence of the Event Rate on Discrimination Abilities of Bankruptcy Predict...
Influence of the Event Rate on Discrimination Abilities of Bankruptcy Predict...Influence of the Event Rate on Discrimination Abilities of Bankruptcy Predict...
Influence of the Event Rate on Discrimination Abilities of Bankruptcy Predict...Lili Zhang
 
Chapter 6 data analysis iec11
Chapter 6 data analysis iec11Chapter 6 data analysis iec11
Chapter 6 data analysis iec11Ho Cao Viet
 
Presentation of Project and Critique.pptx
Presentation of Project and Critique.pptxPresentation of Project and Critique.pptx
Presentation of Project and Critique.pptxBillyMoses1
 
Machine learning and linear regression programming
Machine learning and linear regression programmingMachine learning and linear regression programming
Machine learning and linear regression programmingSoumya Mukherjee
 
Towards Evaluating Size Reduction Techniques for Software Model Checking
Towards Evaluating Size Reduction Techniques for Software Model CheckingTowards Evaluating Size Reduction Techniques for Software Model Checking
Towards Evaluating Size Reduction Techniques for Software Model CheckingAkos Hajdu
 
1440 track 2 boire_using our laptop
1440 track 2 boire_using our laptop1440 track 2 boire_using our laptop
1440 track 2 boire_using our laptopRising Media, Inc.
 
Descriptive Statistics
Descriptive StatisticsDescriptive Statistics
Descriptive StatisticsCIToolkit
 
KNOLX_Data_preprocessing
KNOLX_Data_preprocessingKNOLX_Data_preprocessing
KNOLX_Data_preprocessingKnoldus Inc.
 
Six Sigma Dfss Application In Data Accarucy
Six Sigma Dfss Application In Data AccarucySix Sigma Dfss Application In Data Accarucy
Six Sigma Dfss Application In Data Accarucyxyhfun
 
Week 12 Dimensionality Reduction Bagian 1
Week 12 Dimensionality Reduction Bagian 1Week 12 Dimensionality Reduction Bagian 1
Week 12 Dimensionality Reduction Bagian 1khairulhuda242
 
Terminological cluster trees for Disjointness Axiom Discovery
Terminological cluster trees for Disjointness Axiom DiscoveryTerminological cluster trees for Disjointness Axiom Discovery
Terminological cluster trees for Disjointness Axiom DiscoveryGiuseppe Rizzo
 
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...Edge AI and Vision Alliance
 
Heuristic design of experiments w meta gradient search
Heuristic design of experiments w meta gradient searchHeuristic design of experiments w meta gradient search
Heuristic design of experiments w meta gradient searchGreg Makowski
 
The Machine Learning Workflow with Azure
The Machine Learning Workflow with AzureThe Machine Learning Workflow with Azure
The Machine Learning Workflow with AzureIvo Andreev
 

Semelhante a Feature Engineering (20)

SPC Training by D&H Engineers
SPC Training by D&H EngineersSPC Training by D&H Engineers
SPC Training by D&H Engineers
 
Data_Preparation.pptx
Data_Preparation.pptxData_Preparation.pptx
Data_Preparation.pptx
 
Influence of the Event Rate on Discrimination Abilities of Bankruptcy Predict...
Influence of the Event Rate on Discrimination Abilities of Bankruptcy Predict...Influence of the Event Rate on Discrimination Abilities of Bankruptcy Predict...
Influence of the Event Rate on Discrimination Abilities of Bankruptcy Predict...
 
Chapter 6 data analysis iec11
Chapter 6 data analysis iec11Chapter 6 data analysis iec11
Chapter 6 data analysis iec11
 
Presentation of Project and Critique.pptx
Presentation of Project and Critique.pptxPresentation of Project and Critique.pptx
Presentation of Project and Critique.pptx
 
Machine learning and linear regression programming
Machine learning and linear regression programmingMachine learning and linear regression programming
Machine learning and linear regression programming
 
Towards Evaluating Size Reduction Techniques for Software Model Checking
Towards Evaluating Size Reduction Techniques for Software Model CheckingTowards Evaluating Size Reduction Techniques for Software Model Checking
Towards Evaluating Size Reduction Techniques for Software Model Checking
 
1440 track 2 boire_using our laptop
1440 track 2 boire_using our laptop1440 track 2 boire_using our laptop
1440 track 2 boire_using our laptop
 
report
reportreport
report
 
Descriptive Statistics
Descriptive StatisticsDescriptive Statistics
Descriptive Statistics
 
ML-Unit-4.pdf
ML-Unit-4.pdfML-Unit-4.pdf
ML-Unit-4.pdf
 
KNOLX_Data_preprocessing
KNOLX_Data_preprocessingKNOLX_Data_preprocessing
KNOLX_Data_preprocessing
 
Six Sigma Dfss Application In Data Accarucy
Six Sigma Dfss Application In Data AccarucySix Sigma Dfss Application In Data Accarucy
Six Sigma Dfss Application In Data Accarucy
 
Week 12 Dimensionality Reduction Bagian 1
Week 12 Dimensionality Reduction Bagian 1Week 12 Dimensionality Reduction Bagian 1
Week 12 Dimensionality Reduction Bagian 1
 
Terminological cluster trees for Disjointness Axiom Discovery
Terminological cluster trees for Disjointness Axiom DiscoveryTerminological cluster trees for Disjointness Axiom Discovery
Terminological cluster trees for Disjointness Axiom Discovery
 
07 learning
07 learning07 learning
07 learning
 
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
“Deep Neural Network Training: Diagnosing Problems and Implementing Solutions...
 
Heuristic design of experiments w meta gradient search
Heuristic design of experiments w meta gradient searchHeuristic design of experiments w meta gradient search
Heuristic design of experiments w meta gradient search
 
The Machine Learning Workflow with Azure
The Machine Learning Workflow with AzureThe Machine Learning Workflow with Azure
The Machine Learning Workflow with Azure
 
ictir2016
ictir2016ictir2016
ictir2016
 

Mais de odsc

Understanding the Chief Data Officer
Understanding the Chief Data Officer Understanding the Chief Data Officer
Understanding the Chief Data Officer odsc
 
Machine-In-The-Loop for Knowledge Discovery
Machine-In-The-Loop for Knowledge DiscoveryMachine-In-The-Loop for Knowledge Discovery
Machine-In-The-Loop for Knowledge Discoveryodsc
 
API Driven Development
API Driven Development API Driven Development
API Driven Development odsc
 
Mobile technology Usage by Humanitarian Programs: A Metadata Analysis
Mobile technology Usage by Humanitarian Programs: A Metadata AnalysisMobile technology Usage by Humanitarian Programs: A Metadata Analysis
Mobile technology Usage by Humanitarian Programs: A Metadata Analysisodsc
 
Productionizing Deep Learning From the Ground Up
Productionizing Deep Learning From the Ground UpProductionizing Deep Learning From the Ground Up
Productionizing Deep Learning From the Ground Upodsc
 
Big Data Infrastructure: Introduction to Hadoop with MapReduce, Pig, and Hive
Big Data Infrastructure: Introduction to Hadoop with MapReduce, Pig, and HiveBig Data Infrastructure: Introduction to Hadoop with MapReduce, Pig, and Hive
Big Data Infrastructure: Introduction to Hadoop with MapReduce, Pig, and Hiveodsc
 
Think Breadth, Not Depth
Think Breadth, Not DepthThink Breadth, Not Depth
Think Breadth, Not Depthodsc
 
Data Science at Dow Jones: Monetizing Data, News and Information
Data Science at Dow Jones: Monetizing Data, News and InformationData Science at Dow Jones: Monetizing Data, News and Information
Data Science at Dow Jones: Monetizing Data, News and Informationodsc
 
Spark, Python and Parquet
Spark, Python and Parquet Spark, Python and Parquet
Spark, Python and Parquet odsc
 
Building a Predictive Analytics Solution with Azure ML
Building a Predictive Analytics Solution with Azure MLBuilding a Predictive Analytics Solution with Azure ML
Building a Predictive Analytics Solution with Azure MLodsc
 
Beyond Names
Beyond NamesBeyond Names
Beyond Namesodsc
 
How Woman are Conquering the S&P 500
How Woman are Conquering the S&P 500How Woman are Conquering the S&P 500
How Woman are Conquering the S&P 500odsc
 
Domain Expertise and Unstructured Data
Domain Expertise and Unstructured DataDomain Expertise and Unstructured Data
Domain Expertise and Unstructured Dataodsc
 
Kaggle The Home of Data Science
Kaggle The Home of Data ScienceKaggle The Home of Data Science
Kaggle The Home of Data Scienceodsc
 
Open Source Tools & Data Science Competitions
Open Source Tools & Data Science Competitions Open Source Tools & Data Science Competitions
Open Source Tools & Data Science Competitions odsc
 
Machine Learning with scikit-learn
Machine Learning with scikit-learnMachine Learning with scikit-learn
Machine Learning with scikit-learnodsc
 
Bridging the Gap Between Data and Insight using Open-Source Tools
Bridging the Gap Between Data and Insight using Open-Source ToolsBridging the Gap Between Data and Insight using Open-Source Tools
Bridging the Gap Between Data and Insight using Open-Source Toolsodsc
 
Top 10 Signs of the Textpocalypse
Top 10 Signs of the TextpocalypseTop 10 Signs of the Textpocalypse
Top 10 Signs of the Textpocalypseodsc
 
The Art of Data Science
The Art of Data Science The Art of Data Science
The Art of Data Science odsc
 
Frontiers of Open Data Science Research
Frontiers of Open Data Science ResearchFrontiers of Open Data Science Research
Frontiers of Open Data Science Researchodsc
 

Mais de odsc (20)

Understanding the Chief Data Officer
Understanding the Chief Data Officer Understanding the Chief Data Officer
Understanding the Chief Data Officer
 
Machine-In-The-Loop for Knowledge Discovery
Machine-In-The-Loop for Knowledge DiscoveryMachine-In-The-Loop for Knowledge Discovery
Machine-In-The-Loop for Knowledge Discovery
 
API Driven Development
API Driven Development API Driven Development
API Driven Development
 
Mobile technology Usage by Humanitarian Programs: A Metadata Analysis
Mobile technology Usage by Humanitarian Programs: A Metadata AnalysisMobile technology Usage by Humanitarian Programs: A Metadata Analysis
Mobile technology Usage by Humanitarian Programs: A Metadata Analysis
 
Productionizing Deep Learning From the Ground Up
Productionizing Deep Learning From the Ground UpProductionizing Deep Learning From the Ground Up
Productionizing Deep Learning From the Ground Up
 
Big Data Infrastructure: Introduction to Hadoop with MapReduce, Pig, and Hive
Big Data Infrastructure: Introduction to Hadoop with MapReduce, Pig, and HiveBig Data Infrastructure: Introduction to Hadoop with MapReduce, Pig, and Hive
Big Data Infrastructure: Introduction to Hadoop with MapReduce, Pig, and Hive
 
Think Breadth, Not Depth
Think Breadth, Not DepthThink Breadth, Not Depth
Think Breadth, Not Depth
 
Data Science at Dow Jones: Monetizing Data, News and Information
Data Science at Dow Jones: Monetizing Data, News and InformationData Science at Dow Jones: Monetizing Data, News and Information
Data Science at Dow Jones: Monetizing Data, News and Information
 
Spark, Python and Parquet
Spark, Python and Parquet Spark, Python and Parquet
Spark, Python and Parquet
 
Building a Predictive Analytics Solution with Azure ML
Building a Predictive Analytics Solution with Azure MLBuilding a Predictive Analytics Solution with Azure ML
Building a Predictive Analytics Solution with Azure ML
 
Beyond Names
Beyond NamesBeyond Names
Beyond Names
 
How Woman are Conquering the S&P 500
How Woman are Conquering the S&P 500How Woman are Conquering the S&P 500
How Woman are Conquering the S&P 500
 
Domain Expertise and Unstructured Data
Domain Expertise and Unstructured DataDomain Expertise and Unstructured Data
Domain Expertise and Unstructured Data
 
Kaggle The Home of Data Science
Kaggle The Home of Data ScienceKaggle The Home of Data Science
Kaggle The Home of Data Science
 
Open Source Tools & Data Science Competitions
Open Source Tools & Data Science Competitions Open Source Tools & Data Science Competitions
Open Source Tools & Data Science Competitions
 
Machine Learning with scikit-learn
Machine Learning with scikit-learnMachine Learning with scikit-learn
Machine Learning with scikit-learn
 
Bridging the Gap Between Data and Insight using Open-Source Tools
Bridging the Gap Between Data and Insight using Open-Source ToolsBridging the Gap Between Data and Insight using Open-Source Tools
Bridging the Gap Between Data and Insight using Open-Source Tools
 
Top 10 Signs of the Textpocalypse
Top 10 Signs of the TextpocalypseTop 10 Signs of the Textpocalypse
Top 10 Signs of the Textpocalypse
 
The Art of Data Science
The Art of Data Science The Art of Data Science
The Art of Data Science
 
Frontiers of Open Data Science Research
Frontiers of Open Data Science ResearchFrontiers of Open Data Science Research
Frontiers of Open Data Science Research
 

Último

What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024Stephanie Beckett
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Mattias Andersson
 
SAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxSAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxNavinnSomaal
 
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Mark Simos
 
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo DayH2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo DaySri Ambati
 
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks..."LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...Fwdays
 
DSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningDSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningLars Bell
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek SchlawackFwdays
 
Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Manik S Magar
 
Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 3652toLead Limited
 
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024BookNet Canada
 
WordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your BrandWordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your Brandgvaughan
 
Story boards and shot lists for my a level piece
Story boards and shot lists for my a level pieceStory boards and shot lists for my a level piece
Story boards and shot lists for my a level piececharlottematthew16
 
Search Engine Optimization SEO PDF for 2024.pdf
Search Engine Optimization SEO PDF for 2024.pdfSearch Engine Optimization SEO PDF for 2024.pdf
Search Engine Optimization SEO PDF for 2024.pdfRankYa
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationSlibray Presentation
 
Vertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsVertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsMiki Katsuragi
 
Gen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfGen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfAddepto
 
The Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsThe Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsPixlogix Infotech
 

Último (20)

DMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special EditionDMCC Future of Trade Web3 - Special Edition
DMCC Future of Trade Web3 - Special Edition
 
What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024What's New in Teams Calling, Meetings and Devices March 2024
What's New in Teams Calling, Meetings and Devices March 2024
 
Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?Are Multi-Cloud and Serverless Good or Bad?
Are Multi-Cloud and Serverless Good or Bad?
 
SAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptxSAP Build Work Zone - Overview L2-L3.pptx
SAP Build Work Zone - Overview L2-L3.pptx
 
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
Tampa BSides - Chef's Tour of Microsoft Security Adoption Framework (SAF)
 
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo DayH2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
H2O.ai CEO/Founder: Sri Ambati Keynote at Wells Fargo Day
 
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks..."LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
"LLMs for Python Engineers: Advanced Data Analysis and Semantic Kernel",Oleks...
 
DSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine TuningDSPy a system for AI to Write Prompts and Do Fine Tuning
DSPy a system for AI to Write Prompts and Do Fine Tuning
 
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
"Subclassing and Composition – A Pythonic Tour of Trade-Offs", Hynek Schlawack
 
Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!Anypoint Exchange: It’s Not Just a Repo!
Anypoint Exchange: It’s Not Just a Repo!
 
Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365Ensuring Technical Readiness For Copilot in Microsoft 365
Ensuring Technical Readiness For Copilot in Microsoft 365
 
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptxE-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
E-Vehicle_Hacking_by_Parul Sharma_null_owasp.pptx
 
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
New from BookNet Canada for 2024: BNC CataList - Tech Forum 2024
 
WordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your BrandWordPress Websites for Engineers: Elevate Your Brand
WordPress Websites for Engineers: Elevate Your Brand
 
Story boards and shot lists for my a level piece
Story boards and shot lists for my a level pieceStory boards and shot lists for my a level piece
Story boards and shot lists for my a level piece
 
Search Engine Optimization SEO PDF for 2024.pdf
Search Engine Optimization SEO PDF for 2024.pdfSearch Engine Optimization SEO PDF for 2024.pdf
Search Engine Optimization SEO PDF for 2024.pdf
 
Connect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck PresentationConnect Wave/ connectwave Pitch Deck Presentation
Connect Wave/ connectwave Pitch Deck Presentation
 
Vertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering TipsVertex AI Gemini Prompt Engineering Tips
Vertex AI Gemini Prompt Engineering Tips
 
Gen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdfGen AI in Business - Global Trends Report 2024.pdf
Gen AI in Business - Global Trends Report 2024.pdf
 
The Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and ConsThe Ultimate Guide to Choosing WordPress Pros and Cons
The Ultimate Guide to Choosing WordPress Pros and Cons
 

Feature Engineering

  • 1. FEATURE ENGINEERING David Epstein O P E N D A T A S C I E N C E C O N F E R E N C E_ BOSTON 2015 @opendatasci
  • 2. FEATURE ENGINEERING David Epstein Senior Data Scientist, Socure Open Data Science Conference, Boston, MA May 30-31, 2015 #ODSC, @opendatasci
  • 3. Talk Outline • What is feature engineering? • Limits on number of features • How to select a “good” set of features • Standard FE techniques • TL;DR: As we get better and better models, focus shifts to what we put into them • FE interacts with other key areas of DS
  • 4. Feature Engineering • (My) Definition: Transforming data to create model inputs. Raw Data Data Cleaning Feature Engineering Model Building Pre-Processing Data Workflow
  • 5. Examples from Kaggle Competitions Netflix Titanic Portuguese Taxis
  • 6. How does it work? Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 0.29879 0.48994 0.61 0.559 x -0.00923 0.10256 -0.09 0.930 x2 0.98803 0.03672 26.91 3.92e-09 *** --- Residual standard error: 1.076 on 8 degrees of freedom Multiple R-squared: 0.9891, Adjusted R-squared: 0.9863 F-statistic: 362 on 2 and 8 DF, p-value: 1.427e-08 Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 10.17912 2.92472 3.48 0.00693 ** x -0.00923 0.92488 -0.01 0.99225 --- Residual standard error: 9.7 on 9 degrees of freedom Multiple R-squared: 1.107e-05, Adjusted R-squared: -0.11 F-statistic: 9.96e-05 on 1 and 9 DF, p-value: 0.9923
  • 7. Features are engineered everywhere Things to be explained/measured/predicted Finance: EBITDA Baseball: Batting Avg. Politics: Partisan Performance
  • 8. The Big Questions • Seen in this light, FE is ubiquitous (as all truly important concepts are) • Any time you construct an intermediate variable, you’re doing FE • Two questions naturally arise: 1. How do you construct “good” features? 2. What are the limits on this process? • I’ll answer the second one first, because it’s easier….
  • 9. Limits on Feature Engineering • In medical studies, social science, financial analysis, etc., two main problems emerge • Eating up degrees of freedom: relatively small data sets • # of respondents in survey • # of patients in trial • # of elections to Congress • If your data lives in an NxK matrix, you want to make sure that K is small relative to N • Relevance to hypothesis testing, emphasis on explanation • You generally start with an equation defining the relationship between the key independent and dependent variables • Other variables enter your model as controls, not really interested in their functional form
  • 10. Limits on Feature Engineering • In most modern data science applications, neither is an issue • We start with lots of data, and • Care more about prediction than explanation • So why not add in lots of extra variables? • Think of your data not as what goes into your model, but a starting point for the creation of new variables, which can then be combined…
  • 11. Limits on Feature Engineering • First, adding many correlated predictors can decrease model performance • Adding an x4 term to above example actually reduces model fit: • More variables make models less interpretable • Models have to be generalizable to other data • Too much feature engineering can lead to overfitting • Close connection between feature engineering and cross-validation Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) 0.648123 0.628178 1.032 0.336513 x -0.009230 0.103740 -0.089 0.931593 x2 0.866738 0.139087 6.232 0.000432 *** x4 0.004852 0.005361 0.905 0.395572 --- Residual standard error: 1.088 on 7 degrees of freedom Multiple R-squared: 0.9902, Adjusted R-squared: 0.986 F-statistic: 236.1 on 3 and 7 DF, p-value: 2.15e-07
  • 12. How To Select a “Good” Set of Features • This is the open-ended question in the field • Separate (but related to) the question of feature selection – which variables to retain in a model. • You can use some metrics to tell which features are useful, one at a time, like Pearson correlation coefficient • But this can’t tell you which set of features works best together • This is an NP complete problem, clearly too computationally hard • Many new data analysis services include automated feature engineering as part of their packages • But if there are features you want in your model, it’s best to add them in explicitly, rather than depend on these generators.
  • 13. A “Middle Theory” of FE • Start with a reasonable-sized set of features • Include features suggested by domain knowledge • Test these out individually, build from the bottom up • Number and type of features depend on model used • Can include more features if models does some feature selection • Lasso regression, e.g., logit with ||L1|| regularization (but not ||L2|| ridge) • GBM with backward pruning (but not random forests) • Stepwise regression, with either forward or backward feature selection • Some models are invariant to monotonic variable transformations • Tree-based approaches divide variables into two groups at each branch • So, no perfect answer. But there are some standard techniques every data scientist should have in their bag of tricks.
  • 14. Non-numeric to numeric 1. Count # of times each value appears Zip Code Count 10024 4 63105 2 94304 1 06443 3 10024 4 63105 2 06443 3 10024 4 10024 4 06443 3
  • 15. Non-numeric to numeric 2. One-hot encoding Religion Catholic Protestant Jewish Muslim Catholic 1 0 0 0 Muslim 0 0 0 1 Jewish 0 0 1 0 Protestant 0 1 0 0 Catholic 1 0 0 0 Catholic 1 0 0 0 Jewish 0 0 1 0 Protestant 0 1 0 0 Muslim 0 0 0 1 Protestant 0 1 0 0
  • 16. Non-numeric to numeric 3. The “hash trick” string h(string) “The” 36 “quick” 8 “brown” 92 “fox” 14 “jumps” 75 “over” 25 “the” 36 “lazy” 44 “dog” 21
  • 17. Non-numeric to numeric 4. Leave-one-out encoding
  • 19. Two-variable combinations 1. Add: Sum similar-scaled variables Q1 Q2 Q3 Q4 Total 33 88 51 81 251 11 11 72 30 124 15 36 70 55 176 70 82 8 50 209 99 56 35 86 276 7 20 10 71 107 65 0 25 74 164 96 25 2 89 211 60 29 56 92 238 63 50 96 61 269
  • 20. Two-variable combinations 2. Subtract: Difference relative to baseline ViewerID MovieID Date Rating Days Since First Rating 44972004 8825 1/1/13 5 0 44972004 0471 2/1/13 4 31 44972004 3816 3/1/13 5 59 44972004 8243 4/1/13 3 90 44972004 2855 5/1/13 5 120 44972004 9923 6/1/13 2 151 44972004 1023 7/1/13 4 181 44972004 8306 8/1/13 3 212 44972004 2771 9/1/13 2 243 44972004 5281 10/1/13 2 273
  • 21. Two-variable combinations 3. Multiply: Interactive effects ViewerID Country Domestic DSFR Dom*DSFR 8825 US 1 38 38 0471 CA 0 277 0 3816 FR 0 187 0 8243 US 1 33 33 2855 US 1 87 87 9923 GB 0 42 0 1023 IT 0 192 0 8306 CA 0 365 0 2771 US 1 505 505 5281 FI 0 49 0
  • 22. Two-variable combinations 4. Divide: Scaling/Normalizing Country GDP Population GDP/Capit a US 159849 275 581 CA 731812 111 6593 FR 826320 90 9181 IT 573494 80 7169 IR 609223 22 27692 GB 717673 60 11961 NE 605257 15 40350 MX 687944 124 5548 RU 203319 402 506 FI 744983 40 18625
  • 23. Multivariate/Model-based Methods 1. PCA/Factor Analysis/Clustering: Dimension reduction
  • 24. Multivariate/Model-based Methods 2. Model Stacking: Outputs of one model are inputs to the next • Do this, e.g., on half the data and use as input to the other half, and vice-versa (Figure from Owen Zhang)
  • 25. Conclusion • Data science grew over the last decade due to improved modeling algorithms, better cross- validation procedures • Now we have a number of good models, can get at problem from lots of different angles • Most improvement will come from thinking carefully about what we put into our models = Feature Engineering • This can be (semi-) automated, but it’s still one of the true arts of the profession • Domain knowledge remains very important in practice

Notas do Editor

  1. Seen in this light, FE is ubiquitous (as all truly important concepts are) Any time you construct an intermediate variable, you’re doing FE