Description of Reproduction

extract_news.py

extract news from csv into news_folder

data_process.py

create pickle of daily news for each company ( {str : date : [str : newsID]} )

bind.py

bind news with company and delete news without binding

vectorize.py

doc2vec method for every news

make_stock_trend.py

create pickle of daily trend for each company ( {str : date : int : trend (-1,0,1)} )

make_dataset.py

create data for training and testing

han.py

model training

Python 3.7 with Keras