zhuzhenping / using-deep-learning-neural-networks-and-candlestick-chart-representation-to-predict-stock-market

Stock-Market-Predcition-using-ResNet

1. Prepare Environment

run init.sh to create a virtual python environment and install the dependencies

$ bash init.sh

if you don't want use virtual environment, you can install requirement libraries with :

$ pip install -r requirements.txt

Highly recomended using virtual environment.

2. Prepare dataset

To download data, we provide 2 source, yahoo and tiingo (yahoo by default). We can read a list of stock market and run it. Example, we want to download and preprocess all stock market in tw50.csv with 20 period days and produce 50x50 image dimension.

$ python runallfromlist.py tw50.csv 20 50

Generate the final dataset. Example, we want to generate a final dataset from tw50 with 20 period days and 50 dimension.

$ python generatebigdata.py dataset 20_50 bigdata_20_50

3. Build the model

We can run build model with default parameter.

$ python myDeepCNN.py -i dataset/bigdata_20_50