Agile Experiments in Machine Learning with F#
summary
Shows how the team approached iterative model development for the Kaggle Home Depot competition using F#, describing the experimental workflow, the tooling and harness they built to create and run many ML experiments quickly, and the specific challenges encountered compared to traditional software development and which practices could be adapted between the two contexts.
AI-written summary · about
chapters (18)
- 0:00Intro
- 0:17About me
- 4:11Code on GitHub
- 5:13Kaggle Home Depot
- 15:33What we want
- 15:51What we really mean
- 19:40Overall process
- 19:58ML: experiments in code
- 20:56Learning with Algorithms
- 31:45What are the problems?
- 32:55Core model
- 34:34Catalog of Features
- 38:51Did it work?
- 43:00Dataset normalization
- 46:00Pre-processing pipeline
- 46:37Lesson learnt
- 46:56Tension
- 48:47General
related talks
51mAgile experiments in Machine Learning with F#
01h 02mMachine Learning with F#, Redux
43mAgile Strategies for High-Impact Data Science & Machine Learning Projects
39mDecision Making Based on Machine Learning at Outfittery
26mML Product Experiments at Scale
59mA developers guide to Machine Learning
57mGood Practices for Applied Machine Learning
01h 03mExperimentation in Data Science