Timeband

Timeband is an LSTM-GAN based model for simultaneous detection and correction of missing and outliers in multivariate time series data.

Step 1. Setting up configuration
Step 2. Prepare Time series Dataset
Step 3. Process the Time series Dataset
    - Preprocess with Normalizing / Scaling / ...
    - Process with Holdout / Windowing / ...  
Step 4. Prepare Input/Output Data Structure
    Real Dataset => Encoder => Context Space => Decoder => Target Dataset
Step 5. Train the LSTM-GAN based model
Step 6. Evaluate the models
Step 7. Get the outputs

---

Installation

Requirements

• Nvidia device with CUDA

• Python 3.8+

  Download here

• PyTorch 1.9+

  Download here

• Numpy / pandas / ...

  Download by pip install -r requirements.txt

Code installation

1. Create a virtual environment.

   python -m venv .venv

2. Install PyTorch

   pip install torch==1.9.0 
   or 
   pip install torch==1.9.1+cu111 -f https://download.pytorch.org/whl/torch_stable.html

3. Get Timeband

   git clone https://github.com/handal95/Timeband.git

4. Install Python packages

   pip install -r requirements.txt 

5. Set Default configuration

   • copy config.sample.json to config.json