LSTM Neural Networks

Step by step instructions for setting up a conda development environment:

1. Download this repository using git clone or similar method:
   git clone https://github.com/EPantelaios/LSTM-Neural-Networks.git
2. Install Anaconda from official source based on your OS:
   https://www.anaconda.com/products/individual
3. Run this command to setup the conda environment:
   conda env create --file project3_env.yml
4. Run programs in conda environment with:
   
   • python forecast.py -d nasdaq2007_17.csv -n 10
     
   • python detect.py -d nasdaq2007_17.csv -n 10 -m 0.5
     
   • python reduce.py -d nasdaq_input.csv -q nasdaq_query.csv

In the respective folders there are the following implementations:

1. Using LSTM neural network for foracasting timeseries
2. Using LSTM neural network for timeseries anomaly detection
3. Using convolutional neural network autoencoder for dimensionality reduction of timeseries

Quick Notes:

• There are pre-trained models for each folder and are ready to be used
• In this example NASDAQ share prices are used as timeseries.
• Both .ipynb and .py files are included.