arpithaananth / time-series-forecasting-with-ml-rnn-lstm Goto Github PK

• To identify trends & insights from the data
• To use traditional multivariate ML methods for forecasting
• To implement RNN-LSTM algorithm to forecast the trends
• To compare the results of ML methods vis-a-vis RNN-LSTM algorithm

Stock Market Dataset has been obtained from Data World.

• Date Pulled: 08/29/2017 11:30pm
• Total Snapshots per Index: 2,083
• First Date Captured: September 2, 1977
• Last Date Captured: August 29, 2017

The stock market is a large indicator of the health of the economy. Understanding the stock market, helps in predicting its trends. For this study, Nasdaq, Dow Jones, and S&P 500 market indexes are examined.

The Dow Jones Industrial Average (DJIA) is a widely-watched benchmark index in the U.S. for blue-chip stocks. DJIA is a price-weighted index that tracks 30 large, publicly-owned companies trading on the New York Stock Exchange and the NASDAQ. The index was created by Charles Dow in 1896 to serve as a proxy for the broader U.S. economy.

The Nasdaq Composite Index is a large market-cap-weighted index of more than 2,500 stocks, American depositary receipts (ADRs), and real estate investment trusts (REITs), among others.

The S&P 500 Index or the Standard & Poor's 500 Index is a market-capitalization-weighted index of the 500 largest U.S. publicly traded companies. The index is widely regarded as the best gauge of large-cap U.S. equities.

The insights are found from DJIA's 30 largest publicly owned companies.

It is found that Boeing tops the chart , followed by Goldman Sachs

Walmart has the highest number of employees

Apple has the highest volume, followed by General Electric

Amount here referes to Price * Volume, Apple has the highest amount followed by JPMorgan Chase

• As part of forcasting VAR, VMA, VARMA & VARMAX were implemented
• The Dataset was integrated to the first order to bring stationarity
• Augumented Dickey Fuller, Granger Causality test was done to justify the approach
• Johansen Co-integration test revealed that long term forecasting can be made with the data

• Data was made stationary & was scaled to suit the RNN-LSTM model
• 70:30 was the train-test split
• With 70 units RNN-LSTM model was created

RNN-LSTM gives the best forecast, followed by VAR Model.