liyunchen / CSEDMDataChallenge

2nd CSEDM Data Challenge Sample Code

This repository contains sample code for the 2nd CSEDM Data Challenge. It contains:

• naive_model.ipynb: Basic sample code for Track 1 and 2 of the competition, using simple features.
• code_feature_model.ipynb: A more advanced example of a predictive model using code features for Track 1.
• progsnap2.py: A basic library for loading data from ProgSnap2 datasets.
• CodaLabPackages: The CodaLab packages (minus data and solutions) used to create the competition.
• preprocess.ipynb and compare_semesters.ipynb: The original code used for preprocessing and data labeling and splitting (provided for transparency). Note: This code will not run, since you do not have access to the full datasets used to generate them.

Setup

To run the code in the .ipynb files, you will first need data! Links to the data can be found on the 2nd CSEDM Data Challenge description page.

You should create the following folder structure:

data/Release
    S19: This folder comes from the S19_Release.zip file you downloaded.
        Test
        Train
    F19
        Test: This folder comes from the F19_Release_Test.zip file you downloaded.
        Train: This folder comes from the F19_Release_Train.zip file you downloaded*.

*Note: You may not yet have access to the F19/Train dataset yet. It will be released at the start of the Within-Semester phase of the competition.

Each model is written as a jupyter notebook. To run them, make sure you have Jupyter installed.

You may need to pip install some required packages:

• pandas
• numpy
• matplotlib
• sklearn