Harmon Foods was experiencing significant challenges with forecasting in the sales of Treat, their ready-to-eat breakfast cereal. Inaccuracy in forecasting the sales for the Treat created a big problem that led to costly production changes resulting in lower profitability. The company faces a variation of 50% to 200% in comparison with its actual sales. The company was also facing inventory problems, increase in their cost of production, and the profitability of the company became low. Moreover, the advertisement and the marketing cost is also increased as a result of poor forecasting techniques adopted by the company.
To mitigate these issues, the company must develop a robust sales forecasting methodology to accurately predict future sales and make informed business decisions.
The objective of this project is to develop and estimate a forecasting model for Harmon Foods. The forecasting model will consider various factors such as time trends, seasonality, consumer packs, and dealer allowance. Additionally, we will conduct a Durbin-Watson test to check for autocorrelation in the residuals of the model. Finally, we will generate a forecast along with prediction intervals to provide insights into future case shipments for Harmon Foods.
We collected historical data on case shipments from Harmon Foods. The dataset includes information on case shipments (cs), month (mo), consumer packs (cp, cp.1, cp.2), and other relevant variables such as dealer allowance (da, da.1, da.2) and seasonal index (si), where cp.1, cp.2, da.1, da.2 represents the lags of the consumer packs and dealer allowance variables.
We preprocessed the data by renaming column names for better readability and ensuring consistency in data format.
We conducted exploratory data analysis to understand the distribution of case shipments over time, identify any trends or seasonality in the data, and explore the relationship between case shipments and other variables such as consumer packs and promotions.
We developed a forecasting model using linear regression. The model considers time trends, seasonality, promotions, and lags for promotions as predictor variables. We fitted the model to the historical data and assessed its performance using various metrics.
The resulting regression equation is as follows:
Case Shipments / Sales = -88190.00 + 979.80 * mo + 0.42 * cp - 0.21 * cp.1 - 0.01 * cp.2 + 0.07 * da + 0.00 * da.1 - 0.02 * da.2 + 3858.00 * si
We evaluated the forecasting model by analyzing the goodness-of-fit statistics, including the coefficient of determination (R-squared), standard errors, t-values, and p-values. Additionally, we conducted a Durbin-Watson test to check for autocorrelation in the residuals of the model.
Durbin-Watson test data: model DW = 2.3029, p-value = 0.3973 alternative hypothesis: true autocorrelation is not 0
Using the fitted model, we generated forecasts for future case shipments. We also computed prediction intervals to quantify the uncertainty associated with the forecasts.
| Point Forecast | Lo 80 | Hi 80 | Lo 95 | Hi 95 |
|---|---|---|---|---|
| 1 | 496,859.5 | 446,716.4 | 547,002.7 | 419,058.7 |
Through this project, we have developed and estimated a forecasting model for Harmon Foods, considering various factors influencing case shipments. The model provides insights into future case shipments, helping Harmon Foods make informed decisions about inventory management, resource allocation, and strategic planning.