Time Series LSTM

Long Short Term Memory (LSTM) Model for Time Series Forecasting

  Home    Method   About Us 

Back

Description
The LSTM (Long Short-Term Memory) model is a Recurrent Neural Network (RNN) based architecture that is widely used for time series forecasting. Min-Max transformation has been used for data preparation. Here, we have used one LSTM layer as a simple LSTM model and a Dense layer is used as the output layer. Then, compile the model using the loss function, optimizer and metrics. This package is based on Keras and TensorFlow modules and the algorithm of Paul and Garai (2021) .

Methodology

Hochreiter and Schmidhuber introduced the Long Short Term Memory (LSTM) neural network in 1997 to include the benefits of addressing long-term data dependencies. Because of these characteristics, the LSTM model was very practical for financial high-frequency time series data. LSTM model is developed to solve the problems of recurrent neural network (RNN) models, gradient expansion, and gradient disappearance. LSTM model has three memory modules: input gate, output gate and forget gate. The main functions of these three gates are retaining important information and discarding irrelevant information from the units. The LSTM network has been modified and simplified as “TS-LSTM” model for agricultural time series forecasting.

Function Used

ts.lstm( ts, tsLag = 2, xregLag = 0, Epochs = 5, SplitRatio = 0.8 )

Arguments

ts: Time series data
tsLag: Lag of time series data
LSTMUnits: Number of unit in LSTM layer
Epochs: Number of epochs
SplitRatio: Training and testing data split ratio

References
  1. Deep learning technique for forecasting the price of cauliflower.
    Paul RK, Yeasin M, Kumar P, Paul AK, Roy HS.
    Current Science, 2023, 124(9), 1065-1073.  Get access article here
  2. TSLSTM: Long Short Term Memory (LSTM) Model for Time Series Forecasting
    Paul and Yeasin
    R-package, 2023   Get access article here
© Developed by Prakash Kumar,Senior Scientist, ICAR-IASRI, New DELHI-110012, 2026