Description

Code for ICML 2024 paper

Learning from Streaming Data when Users Choose

Jinyan Su, Sarah Dean

Instructions

Create directory ./dataset and ./results.

Data preprocess

First, download MovieLens 10M Dataset into to the ./dataset directory and unzip it.

Then, run

python movieLens_data_preparation.py

to preprocess the MovieLens 10M dataset

MovieLens 10M dataset result

We use the $n=3$(number of service providers), and $\eta_c=1$ and $\zeta=0, 0.2, 0.5, 0.8, 1$, run the following python file to store the result in ./result file

python movieLens.py

Census Data result

python census.py

After storing the results in the ./results directory, the plot of the convergence of the loss and the iterates are shown in plot.ipynb.