README.md

Student Dropout Prediction using 1D CNN-LSTM with Variational Autoencoder Oversampling

Official implementation of our 2022 IEEE LA-CCI paper Student Dropout Prediction using 1D CNN-LSTM with Variational Autoencoder Oversampling by Eduarda C. Coppo, Rhuan S. Caetano, Leandro M. de Lima and Renato A. Krohling.

Install

conda install numpy pandas matplotlib tensorflow==2.2.0 plotly tabulate scikit-learn seaborn keras ipython jupyter

pip install optuna

Enviroment config

conda create --name <env> --file env.yml

Autoencoer config

In file src\imbalanced-sequence-classification-master\utils\config.py

TIMESTEPS: sequence lenght

TIMESTEPS = 4

DATA_DIM: number of features

DATA_DIM = 26

NUM_CLASSES: number of classes

NUM_CLASSES = 2

Acknowledgement

We borrowed and modified code from "Autoencoders and Generative Adversarial Networks for Anomaly Detection for Sequences" by Stephanie Ger and Diego Klabjan. We would like to expresse our gratitdue for the authors of these repositeries.