SlideShare ist ein Scribd-Unternehmen logo
1 von 31
Downloaden Sie, um offline zu lesen
ML Drift
Identifying Issues Before You Have a Problem
Amy E. Hodler
Dec 2021
The fundamental
assumption in any machine
learning model is that the
data and logic used actually
mimics the real-world.
The fundamental
assumption in any machine
learning model is that the
data and logic used actually
mimics the real-world.
Machine Learning Model Drift
The fundamental
assumption in any machine
learning model is that the
data and logic used actually
mimics the real-world. Over time, our ML models
make worse predictions.
Also called “decay”.
Amy E. Hodler
Evangelist, Responsible AI
Fiddler
@amyhodler Responsible AI, Data Science, and Graphs
Building Trust into AI
BEST
PAPER
KEY PARTNERSHIPS
TECHNOLOGY
PIONEER 2020
CB INSIGHTS MOST PROMISING
AI COMPANIES 2021
ENTERPRISE AI GOVERNANCE AND
ETHICAL RESPONSE 2019
We had a model drift over the weekend that
cost $500,000.”
— Chief Data Scientist
When something goes wrong, it takes our data
scientist 2 weeks to troubleshoot the problem.”
— Data Science Director
Don’t Get Too Caught Up In Terminology
ML Drift, Model Drift, Model Decay, Prediction
Drift = Your predictions are getting worse
Experience, types, causes, and indicators of
drift are sometimes used together, overlap,
and don’t have direct mappings to each
other.
Multiple types of drift can happen at the
same time.
How We Experience ML Drift
Not really drift
but can appear
to be
Image: KD nuggets The ravages of concept drift
Key Types of Drift
Concept Drift
Training data with
decision boundary
● Reality/behavioral
change
● Relationships
change, not the
input
P(Y|X)
Probability of
y output
given x input
P(Y|X) Changes
Image: Don’t let your model’s quality drift away by Michał Oleszak
Key Types of Drift
Concept Drift
Training data with
decision boundary
● Reality/behavioral
change
● Relationships
change, not the
input
P(Y|X)
Probability of
y output
given x input
Data Drift*
P(Y|X) Changes
● Data changes
● Fundamental
relationships do
not change
Image: Don’t let your model’s quality drift away by Michał Oleszak
Key Types of Drift
Concept Drift
Training data with
decision boundary
● Reality/behavioral
change
● Relationships
change, not the
input
P(Y|X)
Probability of
y output
given x input
Label Drift
● Output data shifts
● P(Y) Changes
Data Drift*
P(Y|X) Changes
Feature Drift
● Input data shifts
● P(X) Changes
● Data changes
● Fundamental
relationships do
not change
Image: Don’t let your model’s quality drift away by Michał Oleszak
Key Types of Drift
Virtual Drift
Data changes but
boundary still works
Concept Drift
Training data with
decision boundary
● Reality/behavioral
change
● Relationships
change, not the
input
P(Y|X)
Probability of
y output
given x input
Label Drift
● Output data shifts
● P(Y) Changes
Data Drift*
P(Y|X) Changes
Feature Drift
● Input data shifts
● P(X) Changes
Image: Don’t let your model’s quality drift away by Michał Oleszak
Drift Examples for a Loan Application Model
An income level that was
earlier considered
creditworthy is now
considered riskier.
Concept Drift
*Note that if label and data drift happen together and cancel each other out, there is no concept drift. Otherwise, concept drift will be caused by one of the two, since
they are linked by Bayes equation.
Drift Examples for a Loan Application Model
An income level that was
earlier considered
creditworthy is now
considered riskier.
Concept Drift
A larger proportion of
credit-worthy applications
start showing up.
Label Drift
*Note that if label and data drift happen together and cancel each other out, there is no concept drift. Otherwise, concept drift will be caused by one of the two, since
they are linked by Bayes equation.
Drift Examples for a Loan Application Model
An income level that was
earlier considered
creditworthy is now
considered riskier.
Concept Drift
A larger proportion of
credit-worthy applications
start showing up.
Label Drift
Incomes of most
applicants increase or
decrease. Or you suddenly
get more application from
one region.
Feature Drift
*Note that if label and data drift happen together and cancel each other out, there is no concept drift. Otherwise, concept drift will be caused by one of the two, since
they are linked by Bayes equation.
Triggers of ML Model Drift
● Label or feature distribution
changes e.g. product launch in a
new market
● Concept can change. e.g. a
competitor launching a new
service
May require a new model
Real Data Distribution Change
Triggers of ML Model Drift
● Label or feature distribution
changes e.g. product launch in a
new market
● Concept can change. e.g. a
competitor launching a new
service
May require a new model
● Correct data enters at source but
faulty data engineering. E.g.
debt-to-income values & age
values are swapped in the input.
● Incorrect data enters at source.
E.g., due to a front-end issue, a
website form accepts leaving a
field blank.
Real Data Distribution Change Data Integrity Issues
But Really. . . What’s Really Important?
NOT SURE IF DATA CHANGED
OR REALITY CHANGED
Detect Issues
Analyze Root Cause
Fix It!
Detecting Issues
Performance Monitoring &
Supervised Learning
Data Integrity
Monitoring
Data Drift Monitoring &
Unsupervised Learning
X
Performance Monitoring & Supervised Learning
Works well if you have
ground truth/labels!!
Monitor performance metrics
● Statistical measures
● Accuracy, precision,
FPR, AUC etc.
Supervised learning methods - Ref “A Survey of Concept Drift Adaptation”
● Sequential analysis (SPRT - CUMSUM & PH) - tune alarms on false positives
● Statistical process control (SPC) - rate of change
● Monitoring 2 distributions (ADWIN) - more precise, more overhead
Data Drift Monitoring & Unsupervised Learning
Monitor Statistical Distributions Metrics
● Population Stability Index (PSI)
compares current scoring variable to the
predicted probability from training data
● Kullback–Leibler (KL) divergence measures
the differences of one probability distribution
to a reference probability distribution
● Jensen-Shannon (JS) based on KL, measuring the similarity between two
probability distributions - but is notably symmetric and finite.
● Kolmogorov-Smirnov test (or KS test) quantifies a distance between the
distribution of the sample and the cumulative distribution of the reference
(non-parametric)
Data Drift Monitoring & Unsupervised Learning
Un / Semi-supervised Learning
An overview of unsupervised drift
detection methods
● Can be more accurate
● Online methods look at each
instance (batch methods are
more efficient)
● Most are global oriented
● May miss gradual drift /
sensitivity issues
● Not intuitive for explaining
Data Integrity & Outlier Monitoring
Data errors slowly degrade
performance and can look like real drift
● Missing values
● Range and Type mismatch
● Schema mismatch
Changes in the business (new cataloged
products, revoked pricing, etc.)
Broken data pipeline due to bugs or API
updates
Getting to the Root
Cause
Attribute to drift in features
Account for feature importance
Analyze affected traffic
Root-causing drift in Fiddler
Fix It!
Retrain
new data and/or
relabel old data
Model Mgmt
Archive/schedule model,
ensemble balancing
Adapt/Augment
model behavior, weighting,
biz logic, data collections
Q&A
A Few Resources
Fiddler: ML Model Perf Monitoring platform
XAI Summit
Al Infrastructure Alliance —Nonprofit, independent information
An overview of unsupervised drift detection methods
A Survey of Concept Drift Adaptation
Let’s build trust into AI
amy.hodler@fiddler.ai
@amyhodler
www.fiddler.ai
Fiddler MPM Stack: Deep & Versatile
MONITOR
BUILT-IN EXPLAINABILITY
PLUGGABLE MODEL & DATA INGESTION
Ingest Any
Data Source
Plug into
Any Model
Framework
Connect Via
Fiddler API
APP Custom App
ON-PREM
OR CLOUD
Performance
Model Drift & Bias
Data Integrity & Outliers
ANALYZE
Local & Global Explanations
Bias Detection
Auto-slicing for Performance
CONTROL
Model Inventory
Change & Policy Control
Model Reports

Weitere ähnliche Inhalte

Was ist angesagt?

Using MLOps to Bring ML to Production/The Promise of MLOps
Using MLOps to Bring ML to Production/The Promise of MLOpsUsing MLOps to Bring ML to Production/The Promise of MLOps
Using MLOps to Bring ML to Production/The Promise of MLOps
Weaveworks
 
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Codiax
 
Interpretable Machine Learning Using LIME Framework - Kasia Kulma (PhD), Data...
Interpretable Machine Learning Using LIME Framework - Kasia Kulma (PhD), Data...Interpretable Machine Learning Using LIME Framework - Kasia Kulma (PhD), Data...
Interpretable Machine Learning Using LIME Framework - Kasia Kulma (PhD), Data...
Sri Ambati
 

Was ist angesagt? (20)

Explainable AI
Explainable AIExplainable AI
Explainable AI
 
Drifting Away: Testing ML Models in Production
Drifting Away: Testing ML Models in ProductionDrifting Away: Testing ML Models in Production
Drifting Away: Testing ML Models in Production
 
MLOps Bridging the gap between Data Scientists and Ops.
MLOps Bridging the gap between Data Scientists and Ops.MLOps Bridging the gap between Data Scientists and Ops.
MLOps Bridging the gap between Data Scientists and Ops.
 
Using MLOps to Bring ML to Production/The Promise of MLOps
Using MLOps to Bring ML to Production/The Promise of MLOpsUsing MLOps to Bring ML to Production/The Promise of MLOps
Using MLOps to Bring ML to Production/The Promise of MLOps
 
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
Emeli Dral (Evidently AI) – Analyze it: production monitoring for machine lea...
 
Explainable AI (XAI) - A Perspective
Explainable AI (XAI) - A Perspective Explainable AI (XAI) - A Perspective
Explainable AI (XAI) - A Perspective
 
Explainable AI (XAI)
Explainable AI (XAI)Explainable AI (XAI)
Explainable AI (XAI)
 
Interpretable Machine Learning Using LIME Framework - Kasia Kulma (PhD), Data...
Interpretable Machine Learning Using LIME Framework - Kasia Kulma (PhD), Data...Interpretable Machine Learning Using LIME Framework - Kasia Kulma (PhD), Data...
Interpretable Machine Learning Using LIME Framework - Kasia Kulma (PhD), Data...
 
From Data Science to MLOps
From Data Science to MLOpsFrom Data Science to MLOps
From Data Science to MLOps
 
Automated Machine Learning (Auto ML)
Automated Machine Learning (Auto ML)Automated Machine Learning (Auto ML)
Automated Machine Learning (Auto ML)
 
MLOps.pptx
MLOps.pptxMLOps.pptx
MLOps.pptx
 
Explainable AI - making ML and DL models more interpretable
Explainable AI - making ML and DL models more interpretableExplainable AI - making ML and DL models more interpretable
Explainable AI - making ML and DL models more interpretable
 
AutoML - The Future of AI
AutoML - The Future of AIAutoML - The Future of AI
AutoML - The Future of AI
 
What is MLOps
What is MLOpsWhat is MLOps
What is MLOps
 
Machine Learning
Machine LearningMachine Learning
Machine Learning
 
MLOps Using MLflow
MLOps Using MLflowMLOps Using MLflow
MLOps Using MLflow
 
Anomaly Detection and Spark Implementation - Meetup Presentation.pptx
Anomaly Detection and Spark Implementation - Meetup Presentation.pptxAnomaly Detection and Spark Implementation - Meetup Presentation.pptx
Anomaly Detection and Spark Implementation - Meetup Presentation.pptx
 
Apply MLOps at Scale by H&M
Apply MLOps at Scale by H&MApply MLOps at Scale by H&M
Apply MLOps at Scale by H&M
 
Real World End to End machine Learning Pipeline
Real World End to End machine Learning PipelineReal World End to End machine Learning Pipeline
Real World End to End machine Learning Pipeline
 
Machine Learning
Machine LearningMachine Learning
Machine Learning
 

Ähnlich wie ML Drift - How to find issues before they become problems

Neural Network Model
Neural Network ModelNeural Network Model
Neural Network Model
Eric Esajian
 
Supporting B2Bsales forecasting by machine learning - Mirjana Klajic Borstnar
Supporting B2Bsales forecasting by machine learning - Mirjana Klajic BorstnarSupporting B2Bsales forecasting by machine learning - Mirjana Klajic Borstnar
Supporting B2Bsales forecasting by machine learning - Mirjana Klajic Borstnar
Institute of Contemporary Sciences
 

Ähnlich wie ML Drift - How to find issues before they become problems (20)

Building New Data Ecosystem for Customer Analytics, Strata + Hadoop World, 2016
Building New Data Ecosystem for Customer Analytics, Strata + Hadoop World, 2016Building New Data Ecosystem for Customer Analytics, Strata + Hadoop World, 2016
Building New Data Ecosystem for Customer Analytics, Strata + Hadoop World, 2016
 
How to Use Machine Learning as a Product Manager by Wework PM
 How to Use Machine Learning as a Product Manager by Wework PM How to Use Machine Learning as a Product Manager by Wework PM
How to Use Machine Learning as a Product Manager by Wework PM
 
Neural Network Model
Neural Network ModelNeural Network Model
Neural Network Model
 
Supporting B2Bsales forecasting by machine learning - Mirjana Klajic Borstnar
Supporting B2Bsales forecasting by machine learning - Mirjana Klajic BorstnarSupporting B2Bsales forecasting by machine learning - Mirjana Klajic Borstnar
Supporting B2Bsales forecasting by machine learning - Mirjana Klajic Borstnar
 
BSSML17 - Introduction, Models, Evaluations
BSSML17 - Introduction, Models, EvaluationsBSSML17 - Introduction, Models, Evaluations
BSSML17 - Introduction, Models, Evaluations
 
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
 
C3 w5
C3 w5C3 w5
C3 w5
 
laptop price prediction presentation
laptop price prediction presentationlaptop price prediction presentation
laptop price prediction presentation
 
Machine Learning in the Financial Industry
Machine Learning in the Financial IndustryMachine Learning in the Financial Industry
Machine Learning in the Financial Industry
 
Building a New Platform for Customer Analytics
Building a New Platform for Customer Analytics Building a New Platform for Customer Analytics
Building a New Platform for Customer Analytics
 
Why do the majority of Data Science projects never make it to production?
Why do the majority of Data Science projects never make it to production?Why do the majority of Data Science projects never make it to production?
Why do the majority of Data Science projects never make it to production?
 
Monitoring Models in Production
Monitoring Models in ProductionMonitoring Models in Production
Monitoring Models in Production
 
Model evaluation in the land of deep learning
Model evaluation in the land of deep learningModel evaluation in the land of deep learning
Model evaluation in the land of deep learning
 
Predictive Model and Record Description with Segmented Sensitivity Analysis (...
Predictive Model and Record Description with Segmented Sensitivity Analysis (...Predictive Model and Record Description with Segmented Sensitivity Analysis (...
Predictive Model and Record Description with Segmented Sensitivity Analysis (...
 
[DSC Croatia 22] The Ethos of Process in ML - Leanne Fitzpatrick
[DSC Croatia 22] The Ethos of Process in ML - Leanne Fitzpatrick[DSC Croatia 22] The Ethos of Process in ML - Leanne Fitzpatrick
[DSC Croatia 22] The Ethos of Process in ML - Leanne Fitzpatrick
 
Responsible AI in Industry: Practical Challenges and Lessons Learned
Responsible AI in Industry: Practical Challenges and Lessons LearnedResponsible AI in Industry: Practical Challenges and Lessons Learned
Responsible AI in Industry: Practical Challenges and Lessons Learned
 
Adapting data warehouse architecture to benefit from agile methodologies
Adapting data warehouse architecture to benefit from agile methodologiesAdapting data warehouse architecture to benefit from agile methodologies
Adapting data warehouse architecture to benefit from agile methodologies
 
7 Dangerous Myths DBAs Believe about Data Modeling
7 Dangerous Myths DBAs Believe about Data Modeling7 Dangerous Myths DBAs Believe about Data Modeling
7 Dangerous Myths DBAs Believe about Data Modeling
 
Demand forecasting case study
Demand forecasting case studyDemand forecasting case study
Demand forecasting case study
 
Data Storytelling for Insurers Presentation
Data Storytelling for Insurers PresentationData Storytelling for Insurers Presentation
Data Storytelling for Insurers Presentation
 

Kürzlich hochgeladen

Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and Myths
Joaquim Jorge
 

Kürzlich hochgeladen (20)

MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024MINDCTI Revenue Release Quarter One 2024
MINDCTI Revenue Release Quarter One 2024
 
Tata AIG General Insurance Company - Insurer Innovation Award 2024
Tata AIG General Insurance Company - Insurer Innovation Award 2024Tata AIG General Insurance Company - Insurer Innovation Award 2024
Tata AIG General Insurance Company - Insurer Innovation Award 2024
 
A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?A Year of the Servo Reboot: Where Are We Now?
A Year of the Servo Reboot: Where Are We Now?
 
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
Deploy with confidence: VMware Cloud Foundation 5.1 on next gen Dell PowerEdg...
 
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time AutomationFrom Event to Action: Accelerate Your Decision Making with Real-Time Automation
From Event to Action: Accelerate Your Decision Making with Real-Time Automation
 
Artificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and MythsArtificial Intelligence: Facts and Myths
Artificial Intelligence: Facts and Myths
 
Exploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone ProcessorsExploring the Future Potential of AI-Enabled Smartphone Processors
Exploring the Future Potential of AI-Enabled Smartphone Processors
 
Boost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivityBoost PC performance: How more available memory can improve productivity
Boost PC performance: How more available memory can improve productivity
 
presentation ICT roal in 21st century education
presentation ICT roal in 21st century educationpresentation ICT roal in 21st century education
presentation ICT roal in 21st century education
 
🐬 The future of MySQL is Postgres 🐘
🐬  The future of MySQL is Postgres   🐘🐬  The future of MySQL is Postgres   🐘
🐬 The future of MySQL is Postgres 🐘
 
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
Connector Corner: Accelerate revenue generation using UiPath API-centric busi...
 
The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024The 7 Things I Know About Cyber Security After 25 Years | April 2024
The 7 Things I Know About Cyber Security After 25 Years | April 2024
 
Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024Manulife - Insurer Innovation Award 2024
Manulife - Insurer Innovation Award 2024
 
Artificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : UncertaintyArtificial Intelligence Chap.5 : Uncertainty
Artificial Intelligence Chap.5 : Uncertainty
 
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost SavingRepurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
Repurposing LNG terminals for Hydrogen Ammonia: Feasibility and Cost Saving
 
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
Bajaj Allianz Life Insurance Company - Insurer Innovation Award 2024
 
Partners Life - Insurer Innovation Award 2024
Partners Life - Insurer Innovation Award 2024Partners Life - Insurer Innovation Award 2024
Partners Life - Insurer Innovation Award 2024
 
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdfUnderstanding Discord NSFW Servers A Guide for Responsible Users.pdf
Understanding Discord NSFW Servers A Guide for Responsible Users.pdf
 
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, AdobeApidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
Apidays New York 2024 - Scaling API-first by Ian Reasor and Radu Cotescu, Adobe
 
AWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of TerraformAWS Community Day CPH - Three problems of Terraform
AWS Community Day CPH - Three problems of Terraform
 

ML Drift - How to find issues before they become problems

  • 1. ML Drift Identifying Issues Before You Have a Problem Amy E. Hodler Dec 2021
  • 2. The fundamental assumption in any machine learning model is that the data and logic used actually mimics the real-world.
  • 3. The fundamental assumption in any machine learning model is that the data and logic used actually mimics the real-world.
  • 4. Machine Learning Model Drift The fundamental assumption in any machine learning model is that the data and logic used actually mimics the real-world. Over time, our ML models make worse predictions. Also called “decay”.
  • 5. Amy E. Hodler Evangelist, Responsible AI Fiddler @amyhodler Responsible AI, Data Science, and Graphs
  • 6. Building Trust into AI BEST PAPER KEY PARTNERSHIPS TECHNOLOGY PIONEER 2020 CB INSIGHTS MOST PROMISING AI COMPANIES 2021 ENTERPRISE AI GOVERNANCE AND ETHICAL RESPONSE 2019
  • 7. We had a model drift over the weekend that cost $500,000.” — Chief Data Scientist When something goes wrong, it takes our data scientist 2 weeks to troubleshoot the problem.” — Data Science Director
  • 8. Don’t Get Too Caught Up In Terminology ML Drift, Model Drift, Model Decay, Prediction Drift = Your predictions are getting worse Experience, types, causes, and indicators of drift are sometimes used together, overlap, and don’t have direct mappings to each other. Multiple types of drift can happen at the same time.
  • 9. How We Experience ML Drift Not really drift but can appear to be Image: KD nuggets The ravages of concept drift
  • 10. Key Types of Drift Concept Drift Training data with decision boundary ● Reality/behavioral change ● Relationships change, not the input P(Y|X) Probability of y output given x input P(Y|X) Changes Image: Don’t let your model’s quality drift away by Michał Oleszak
  • 11. Key Types of Drift Concept Drift Training data with decision boundary ● Reality/behavioral change ● Relationships change, not the input P(Y|X) Probability of y output given x input Data Drift* P(Y|X) Changes ● Data changes ● Fundamental relationships do not change Image: Don’t let your model’s quality drift away by Michał Oleszak
  • 12. Key Types of Drift Concept Drift Training data with decision boundary ● Reality/behavioral change ● Relationships change, not the input P(Y|X) Probability of y output given x input Label Drift ● Output data shifts ● P(Y) Changes Data Drift* P(Y|X) Changes Feature Drift ● Input data shifts ● P(X) Changes ● Data changes ● Fundamental relationships do not change Image: Don’t let your model’s quality drift away by Michał Oleszak
  • 13. Key Types of Drift Virtual Drift Data changes but boundary still works Concept Drift Training data with decision boundary ● Reality/behavioral change ● Relationships change, not the input P(Y|X) Probability of y output given x input Label Drift ● Output data shifts ● P(Y) Changes Data Drift* P(Y|X) Changes Feature Drift ● Input data shifts ● P(X) Changes Image: Don’t let your model’s quality drift away by Michał Oleszak
  • 14. Drift Examples for a Loan Application Model An income level that was earlier considered creditworthy is now considered riskier. Concept Drift *Note that if label and data drift happen together and cancel each other out, there is no concept drift. Otherwise, concept drift will be caused by one of the two, since they are linked by Bayes equation.
  • 15. Drift Examples for a Loan Application Model An income level that was earlier considered creditworthy is now considered riskier. Concept Drift A larger proportion of credit-worthy applications start showing up. Label Drift *Note that if label and data drift happen together and cancel each other out, there is no concept drift. Otherwise, concept drift will be caused by one of the two, since they are linked by Bayes equation.
  • 16. Drift Examples for a Loan Application Model An income level that was earlier considered creditworthy is now considered riskier. Concept Drift A larger proportion of credit-worthy applications start showing up. Label Drift Incomes of most applicants increase or decrease. Or you suddenly get more application from one region. Feature Drift *Note that if label and data drift happen together and cancel each other out, there is no concept drift. Otherwise, concept drift will be caused by one of the two, since they are linked by Bayes equation.
  • 17. Triggers of ML Model Drift ● Label or feature distribution changes e.g. product launch in a new market ● Concept can change. e.g. a competitor launching a new service May require a new model Real Data Distribution Change
  • 18. Triggers of ML Model Drift ● Label or feature distribution changes e.g. product launch in a new market ● Concept can change. e.g. a competitor launching a new service May require a new model ● Correct data enters at source but faulty data engineering. E.g. debt-to-income values & age values are swapped in the input. ● Incorrect data enters at source. E.g., due to a front-end issue, a website form accepts leaving a field blank. Real Data Distribution Change Data Integrity Issues
  • 19. But Really. . . What’s Really Important? NOT SURE IF DATA CHANGED OR REALITY CHANGED
  • 20. Detect Issues Analyze Root Cause Fix It!
  • 21. Detecting Issues Performance Monitoring & Supervised Learning Data Integrity Monitoring Data Drift Monitoring & Unsupervised Learning X
  • 22. Performance Monitoring & Supervised Learning Works well if you have ground truth/labels!! Monitor performance metrics ● Statistical measures ● Accuracy, precision, FPR, AUC etc. Supervised learning methods - Ref “A Survey of Concept Drift Adaptation” ● Sequential analysis (SPRT - CUMSUM & PH) - tune alarms on false positives ● Statistical process control (SPC) - rate of change ● Monitoring 2 distributions (ADWIN) - more precise, more overhead
  • 23. Data Drift Monitoring & Unsupervised Learning Monitor Statistical Distributions Metrics ● Population Stability Index (PSI) compares current scoring variable to the predicted probability from training data ● Kullback–Leibler (KL) divergence measures the differences of one probability distribution to a reference probability distribution ● Jensen-Shannon (JS) based on KL, measuring the similarity between two probability distributions - but is notably symmetric and finite. ● Kolmogorov-Smirnov test (or KS test) quantifies a distance between the distribution of the sample and the cumulative distribution of the reference (non-parametric)
  • 24. Data Drift Monitoring & Unsupervised Learning Un / Semi-supervised Learning An overview of unsupervised drift detection methods ● Can be more accurate ● Online methods look at each instance (batch methods are more efficient) ● Most are global oriented ● May miss gradual drift / sensitivity issues ● Not intuitive for explaining
  • 25. Data Integrity & Outlier Monitoring Data errors slowly degrade performance and can look like real drift ● Missing values ● Range and Type mismatch ● Schema mismatch Changes in the business (new cataloged products, revoked pricing, etc.) Broken data pipeline due to bugs or API updates
  • 26. Getting to the Root Cause Attribute to drift in features Account for feature importance Analyze affected traffic Root-causing drift in Fiddler
  • 27. Fix It! Retrain new data and/or relabel old data Model Mgmt Archive/schedule model, ensemble balancing Adapt/Augment model behavior, weighting, biz logic, data collections
  • 28. Q&A
  • 29. A Few Resources Fiddler: ML Model Perf Monitoring platform XAI Summit Al Infrastructure Alliance —Nonprofit, independent information An overview of unsupervised drift detection methods A Survey of Concept Drift Adaptation
  • 30. Let’s build trust into AI amy.hodler@fiddler.ai @amyhodler www.fiddler.ai
  • 31. Fiddler MPM Stack: Deep & Versatile MONITOR BUILT-IN EXPLAINABILITY PLUGGABLE MODEL & DATA INGESTION Ingest Any Data Source Plug into Any Model Framework Connect Via Fiddler API APP Custom App ON-PREM OR CLOUD Performance Model Drift & Bias Data Integrity & Outliers ANALYZE Local & Global Explanations Bias Detection Auto-slicing for Performance CONTROL Model Inventory Change & Policy Control Model Reports