EMPIRICAL ANALYSIS OF SALES DATA USING REGRESSION AND ARIMA MODELS FOR BUSINESS FORECASTING

Abstract

This study provides an empirical foundation for business forecasting and offers practical insights to support decision-makers in effectively allocating advertising budgets. Sales data are analyzed using two advanced statistical models: regression analysis and Auto Regressive Integrated Moving Average (ARIMA). The objective is to examine the relationship between sales and advertising expenditures across television, radio, and internet media, and to forecast future sales patterns. Results from the regression model indicate that television advertising has the strongest impact on sales, followed by radio and online spending. The ARIMA model, selected based on the lowest Akaike information criterion (AIC) and Bayesian Information Criterion (BIC) value, produces reliable forecasts that suggest a steady upward trend in sales. Additionally, a linear trend model (LTM) demonstrates a smooth increase in sales, though accompanied by considerable forecasting uncertainty. Collectively, the results highlight the significant role of advertising expenditures in driving sales performance while acknowledging the inherent challenges of prediction. These findings provide valuable guidance for developing effective marketing strategies and improving sales forecasting practices.

Contribution of the Study

This study contributes by empirically integrating regression and ARIMA models to analyze advertising impacts on sales and forecast future trends. The dual approach provides both explanatory insights into effective media spending and predictive accuracy for sales planning, offering valuable guidance for marketing strategy and business forecasting.