LSTM-SDM: An integrated framework of LSTM implementation for sequential data modeling[Formula presented]

Authors

Hum Nath Bhandari, Roger Williams UniversityFollow
Binod Rimal, Florida Atlantic University
Nawa Raj Pokhrel, Xavier University of Louisiana
Ramchandra Rimal, Middle Tennessee State University
Keshab R. Dahal, Truman State University

Document Type

Article

Publication Title

Software Impacts

Publication Date

11-1-2022

Abstract

LSTM-SDM is a python-based integrated computational framework built on the top of Tensorflow/Keras and written in the Jupyter notebook. It provides several object-oriented functionalities for implementing single layer and multilayer LSTM models for sequential data modeling and time series forecasting. Multiple subroutines are blended to create a conducive user-friendly environment that facilitates data exploration and visualization, normalization and input preparation, hyperparameter tuning, performance evaluations, visualization of results, and statistical analysis. We utilized the LSTM-SDM framework in predicting the stock market index and observed impressive results. The framework can be generalized to solve several other real-world time series problems.

Volume

14

DOI

10.1016/j.simpa.2022.100396

Recommended Citation

Bhandari, H., Rimal, B., Pokhrel, N., Rimal, R., & Dahal, K. (2022). LSTM-SDM: An integrated framework of LSTM implementation for sequential data modeling[Formula presented]. Software Impacts, 14 https://doi.org/10.1016/j.simpa.2022.100396

E-ISSN

26659638