Forecasting sales demand of a manufacturing company. A time series approach

Author

Dzvetera, Tanyaradzwa D.

Title

Forecasting sales demand of a manufacturing company. A time series approach

Abstract

Forecasting sales is crucial for manufacturing firms as it enables the identification of future sales patterns for products. This predictive analysis aids in optimizing retail operations to meet customer demand effectively, anticipate potential losses, and manage inventory levels efficiently. The data was collected from the manufacturing company Corked Spin Investment every month from the year 2017 up to 2023. The study investigates the trend analysis, sustainability, and effectiveness of the time series model in forecasting sales demand, it also compares multilayer Perceptron and SARIMA to show which method is the best in forecasting and lastly, it forecasts sales demand for the year 2024 on monthly bases. This research used a quantitative research design. A literature review is performed to identify a suitable model for forecasting sales demand for corked spin investment. Multilayer Perceptron (MLP) and Seasonal Autoregressive Integrated Moving Average (SARIMA) sales demand forecasting techniques were employed, and the outcomes revealed that Multilayer Perception has the best performance than SARIMA in forecasting Corked Spin Investment sales demand. This was done using R programming. From this research, it is concluded that Multilayer Perception can model nonlinear function compared to SARIMA. Hence MLP is chosen as the ideal model for predicting sales of Corked Spin Investment. The trend of MLP is increasing showing a better performance in the year 2024. It is advisable to use MLP to forecast sales of the company. Further studies should enhance the MLP model predictive power by incorporating additional variables such as macroeconomic indicators and competitor activity

Date

2024

Publisher

BUSE

Keywords

Forecasting

Sales demand

Manufacturing company

Time series approach

Supervisor

Dr Mapuwei