search

nginyc / rafiki

Rafiki is a distributed system that supports training and deployment of machine learning models using AutoML, built with ease-of-use in mind.
Apache License 2.0
36 stars 23 forks source link

add two models for structured data( kaggle Home Credit) #184

Open wxupjack opened 4 years ago

wxupjack commented 4 years ago

This pull request is from NUSSZAI.

We add two models for the Home Credit Default Risk task (can be found at Kaggle and the datasets are also supported, from which we only use the main table application_{train|test}.csv ).

One model bases on LightGBM and the other on a mixed input NN using PyTorch. These two models has been successfully tested.

wxupjack commented 4 years ago

These two files are modified to meet the specifications of tabular classification task, in regard to the internal data loading and processing functions.