Model Selection with Python: An Introduction to Hyper Parameter Tuning
summary
Claim: hyperparameter choice is itself a machine learning problem and can be automated. Uses concrete tools: xgboost model with an objective of mean squared error and evaluation by mean absolute error, tuned via RandomizedSearchCV / random search; notes about 22 hyperparameters and that search is highly parallelizable. Mentions using AutoML services to search hyper and model space together for Nimble’s continuous credit intelligence system and that practical constraints (GPU availability, latency, explainability) decide what model and search approach actually gets deployed.
AI-written summary · about
chapters (25)
- 0:00Introduction
- 1:15About me
- 2:04Building a project
- 3:07Modeling
- 5:15Model Selection
- 5:37Too Many Models
- 6:18Slow Chart
- 6:42Defining the Problem
- 7:19InputOutputs
- 9:16Three Scenarios
- 10:19What is Hyper
- 10:41What is HyperParameter Tuning
- 10:59What is a HyperParameter
- 12:33HyperParameter Tuning Example
- 12:44Python Model Selection
- 15:13Neural Net
- 15:25HalfSoul
- 19:27Output
- 19:51After Dinner Conversation
- 20:05Summary
- 20:30Philosophical Question
- 21:16Framing the Problem
- 21:42Coffee Chat
- 22:04Questions
- 25:25Outro
related talks
25mHyperparameter Tuning and Visualization in Deep Learning
18mBest Practices for Hyperparameter Optimization
18mAlexandra Johnson, Software Engineer, SigOpt, NYC 2017
57mGood Practices for Applied Machine Learning
27mAutomating XGBoost Hyperparameter Tuning with Agentic AI
38mA Guide to Modern Hyperparameters Turning Algorithms- 57mGood Practices for Applied Machine Learning
33mTune your ML models to the highest accuracy with automatic model tuning