Method

Feature importance

Permutation importance, PDP, LIME and model-based importances.

45 cases from 41 companies apply Feature importance. 7 were checked in an online A/B test; 13 report results in production. Includes variants: SHAP.

Typical tasks: Evaluate models, Catch fraud and abuse, Predict churn and lifetime value, Detect anomalies and 12 more. Companies: Netflix, Lyft, Thumbtack, Zomato, Affirm, Ramp and 35 more. Often combined with: SHAP, Gradient boosting, Churn prediction, Random forest, Feature store and 97 more.

PredictPay: From Hackathon Idea to Smarter Payments at Scalefintech2026-10
Putting a Transformer Inside Every Card Authorizationfintech2026-10
Roonomics: How Economics Helped Us to Build a Churn Prediction Modelfood-delivery2026-09
Agentic Feature Engineering: How McAfee Drives Personalization With Agent Bricksgeneral2026-06
Reliance Jio Customer Churn Radar and next best actiontelecom2026-06
ML Scoring of 3DS Authentications in a Payment Systemfintech2026-05
How we scaled ranking with Learn-to-Ranke-commerce2026-02
ML Observability: Bringing Transparency to Payments and Beyondmedia-streaming2025-08
ML in the Oil Industry: Can AI Kill a Well?manufacturing2025-05
Causal Methods for Feature Selectiongeneral2025-05
Virtual Flow Metering: Forecasting Oil Production with Machine Learningmanufacturing2025-05
Autotuning: A comprehensive approach to Time Series Forecastingretail2025-03
Behind HubSpot AI: How Does Prediction Engine Score Millions of CRM Objects Daily?saas-b2b2024-11
Challenges of ML Development in an Expert Domain Requiring High Accuracygeneral2024-11
Short-Term Order Forecasting for Creating Courier Slots at Dark Storesfood-delivery2024-05
Predictive Analysis of Infrastructure Risks on Railway Tracksmanufacturing2024-05
Not by Diagnostics Alone: ML for Designing Therapeutic Proteinshealthtech2024-05
Lyft’s Reinforcement Learning Platformmobility2024-03
Recursive Embedding and Clusteringmedia-streaming2023-12
Product Features: The Key to Increasing Model Efficiency for Account Propensitysaas-b2b2023-10
Explaining Autoencoder Anomalies in Industrial Control Systemsmanufacturing2023-10
Risk Entity Watch - Using Anomaly Detection to Fight Fraudmobility2023-09
Opportunities for Automating Retinal OCT Image Analysishealthtech2023-05
Leaves to Neurons: Using Deep-Cross Networks for Ranking Search Resultsmarketplace2023-05
An ML based approach to proactive advertiser churn preventionadtech2023-05
Building a large scale unsupervised model anomaly detection system — Part 2mobility2023-04
How we built it: Stripe Radarfintech2023-03
Discovering Creative Insights in Promotional Artworkmedia-streaming2023-01
TimeSHAP: Explaining recurrent models through sequence perturbationsbanking2023-01
Evolution of Search Ranking at Thumbtackmarketplace2022-12
Science or Nature? A recommender system for choosing a journal for a papergeneral2022-10
Using mobile sensor data to encourage safer drivingmobility2022-10
Graph Machine Learning at Airbnbmarketplace2022-06
Connecting the Dots: strengthening recommendations for our customers (Part One)food-delivery2022-06
From look-alike to uplift in customer lifecycle modelsbanking2022-06
Transformers for Generalizing User Behavior in Yandex.Taximobility2022-06
Comparison of Interpretation Algorithmsgeneral2022-06
A small BERT towards Large Medical Modelshealthtech2022-06
What is new in ML model interpretabilitygeneral2022-06
Introduction to structure learning and causal discoverygeneral2022-06
SHAP Plots: The Crystal Ball for UI Test Ideasmarketplace2022-04
Beyond Matrix Factorization: Using hybrid features for user-business recommendationsmarketplace2022-04
Building the Model Behind DoorDash's Expansive Merchant Selectionfood-delivery2022-04
The journey to build an explainable AI-driven recommendation system to help scale sales efficiency across LinkedInsaas-b2b2022-04
How Zomato built and deployed a cheap, accurate delivery-partner ETA modelfood-delivery2022-03