Machine Learning simplified for Developers with ML.NET
summary
Uses ML.NET Model Builder to train a text classifier from a CSV of bank transactions, prototypes a working model in about an hour and single runs take ~8–15 seconds per trainer, hits ~85% accuracy after pruning useless columns and gains ~1% by removing date, exposes generated C# code and a PredictionEngine for local consumption, fixes a categorical-hash issue by switching description encoding to featurized text and recovers near-perfect classification for that case, supports offline models and a CLI but needs >100 rows to train.
AI-written summary · about
chapters (22)
- 0:00Introduction
- 2:57Agenda
- 8:27MLNET
- 10:52What is Machine Learning
- 15:07ML Operations
- 24:44What is the process
- 25:50How do we get started
- 26:57Caveats
- 28:27MLNET Model Builder
- 29:26MLNET Trainer
- 29:55MLNET Model Builder Preview
- 30:36Data
- 32:06Demo
- 34:08Getting started
- 34:41When to start training
- 35:56Model Builder
- 42:22Evaluation
- 47:12Remove Expenses
- 49:01New Projects
- 50:23Prediction Engine
- 52:31Troubleshooting
- 56:22Summary
related talks
01h 03mMachine learning for C# developers: Introducing ML
46mMachine Learning for .NET Developers
59mMachine Learning in .NET (ML.NET)
45mMachine Learning for .NET developers
58mExploring Machine Learning for .Net Developers
26mPredicting the future (of equipment) using ML.NET
57mMachine Learning Made Easy – An Introduction to ML.NET
27mUsing C# for Data Science and Machine Learning