A Comparative Study Between Lstm And Arima Models In Forecasting Sales For Simbisa Brands Chicken Inn (2016-2023).

Author

Rumbidzai K Mtetwa

Title

A Comparative Study Between Lstm And Arima Models In Forecasting Sales For Simbisa Brands Chicken Inn (2016-2023).

Abstract

As time goes on, an increasing number of restaurants search for answers to their data-related issues. The comparative analysis of LSTM and ARIMA in sales forecasting for Simbisa Brands from 2016 to 2023 was the focus of this research. The primary goal was to assess the accuracy and dependability of the sales estimates produced. During the 12-month internship at Simbisa Brands, the main aim was to increase sales and fight competition as more and more food outlets providing chicken-based products were being introduced. Using a data analytics method, two algorithms which are ARIMA and LSTM were applied, hundreds of time series data were analysed and external variables were incorporated. Python was utilized as the programming language for evaluating both models. After being put into use, SARIMAX was selected as the best model based on performance matrices (RMSE, MAPE AND MAE). Future developments and suggestions were also noted in light of the findings. It is advised that Simbisa Brands think about establishing additional stores in key areas to take advantage of the steady growth trend and grow their market share, according to the SARIMAX out prediction. It is advised that Simbisa Brands add new menu items to stay up to date with shifting consumer tastes and preferences while upholding the standards of quality and value. SARIMA outperformed both ARIMA and SARIMA models making it the best model for Simbisa Brands to use.

Date

June 2024

Publisher

BUSE

Keywords

ARIMA

machine learning

time series,

forecasting

Supervisor

R T.W. MAPUWEI