Simplify the collection, printing, and saving of model metrics in ML.NET. Take the following line:
var modelMetrics = mlContext.Regression.Evaluate(testSetTransform);
We would take the 'modelMetrics' object and print each property out or parse it out to a file. This is tedious code to write every time a model is created. We would like to simplify this down to a single line to, for instance:
Hackathon Idea
Please fill out this form to submit an idea for the Virtual ML.NET Hackathon
Your name
Brandon Atkinson (github.com/atkinsonbg) Nathaniel Brumbach (github.com/Extrabeefy)
Team name
Practical ML.NET User Group Team
Brief Description
Simplify the collection, printing, and saving of model metrics in ML.NET. Take the following line:
var modelMetrics = mlContext.Regression.Evaluate(testSetTransform);
We would take the 'modelMetrics' object and print each property out or parse it out to a file. This is tedious code to write every time a model is created. We would like to simplify this down to a single line to, for instance:
MLContext.Regression.Metrics()
We also plan to include the permutation feature importance in this output: https://docs.microsoft.com/en-us/dotnet/machine-learning/how-to-guides/explain-machine-learning-model-permutation-feature-importance-ml-net
Other
Are you looking for team members?