experiments

fastSparse

This page contains the source code to reproduce results in our AISTATS 2022 paper:

• Fast Sparse Classification for Generalized Linear and Additive Models

The FICO dataset is publically available. You can request the data from this link.

To convert the original FICO dataset with continuous features into binary features, please refer to our header in this file.

1. Time Comparison Experiment

For the time comparison experiments, please run the following line in your terminal

Rscript run_time.R

2. Solution Quality Experiment of the Entire Regularization Path

For the solution quality experiment, please run the following line in your terminal

Rscript run_baseline.R

The above script contains code to run experiments on both the real datasets and the synthetically generated data.