stock-price-prediction-with-CNN-LSTM

This repository contains the source code for the paper "Stock Price Prediction with CNN-LSTM Network".
It performs next day stock price prediction using Dilated CNN, or LSTM, or the combination of both. It also performs simulated training based on the next day stock price prediction. It is found that a combination of Dilated CNN and LSTM give the most accurate prediciton. We have also compared the Neural Network's performance with classical machine learning methods such as KNN, Random Forest and Linear Regression.

Training and Testing

Training

./train.py # generate default results
./data_loader/configs # location of generated result

Modify Configuration

vim ./data_loader/configs/config1.json
vim ./data_loader/configs/config2.json

For the non-neural methods, simply call their corresponding files:

./KNN.py
./RandomForest.py
./Linear_Regression.py