ayra161 / lstm-timeseries-prediction

LSTM using Keras for Time Series data

LSTM Time Series Prediction

LSTM using Keras to predict the time series data. There are two running files to predict international airline passengers and google stock market. We use 65% of data to train the LSTM model and predict the other 35% of data and compare with real data.

• International airline passengers: Number of passengers for each month in thousands.
• google-stock.csv: Google stock market data from 2012 to 2017

Requirements

Please check the version of libraries I used for this LSTM.

• pandas==0.23.3
• numpy==1.15.4
• tensorflow==1.12.0
• keras==2.2.4
• matplotlib==3.0.2

Reference

• international-airline-passengers.csv: https://www.kaggle.com/andreazzini/international-airline-passengers
• google-stock.csv: https://www.kaggle.com/wogus934/google-stock

Output for passengers_lstm.py.

• Train Score: 0.0016482554761827904
• Test Score: 0.0017052177690008714

Output for google_stock_lstm.py

• Train Score: 0.0022733879506346734
• Test Score: 0.012751589032441776