Reference no: EM133376147
The Grocery Sales Organization (GSO), a well-established chain of grocery stores, has recently hired you as the new executive officer in charge of continuous improvement. In the months to come, you have been asked to help them review their processes in order to become more productive and stay competitive in this highly aggressive market.
After only a few weeks, you have identified several issues that should be investigated and, hopefully, improved. Since the main issue seems to be a drop in client satisfaction, you decide to start investigating there. You quickly realize that the problem most frequently mentioned by clients seems to be that grocery stores frequently run out of products that are on sale. Knowing that good inventory management is highly dependant on demand forecasts, you decide to verify if the company's forecasting methods are adequate by focussing on a particular product to begin with.
1. The forecasting technique currently used to manage inventory at GSO's distribution center (DC) has given the results shown in Table 1. The forecasting model is a 4-week weighted moving average with weights of 0.5, 0.3, 0.1 and 0.1. According to the MAD and TS, is this forecasting model giving adequate forecasts? To answer this question, you must:
a) Illustrate a graph showing two curves: the demand and the forecast
b) Calculate the Mean absolute deviation (MAD)
c) Calculate the Running sum of forecasting error (RSFE)
d) Illustrate the Tracking signal (TS) in a graph also showing the upper and lower limits of +4 and -4.
e) Draw your conclusions based on these elements
Perplexed by this analysis, you decide to investigate further into the data used by the forecasting model. After questioning many employees, you finally discover that the 'demand data' is in fact the quantity of cases shipped from GSO's DC to its stores, based on the shipping manifests. Having the feeling that this could be the source of the problem, you hunt down the data needed to provide a better idea of the actual demand. You find records of the orders placed by the stores and realize that the quantities shipped do not match. Checking with the warehouse manager, you find out that when ordered quantities are too high, they simply ship the quantity on hand at the DC and continue shipping the backorders in the following weeks. You realize that this could indeed be a problem if the extra quantities were needed at the stores who expected a higher demand when the product was offered at a discount.
2) Table 2 shows the ordered quantities per week. Use this data as the 'demand' in the current forecasting method to calculate new forecasts (calculate forecasts starting at period 6). Is the model giving adequate forecasts? Use the same steps as for question 1.
3) You decide to also try an exponential smoothing model with a constant α=0.2 to forecast the ordered quantities per week (Table 2). Start with a forecast for week 6 equal to the average demand over weeks 1 to 5. Use the same steps as for question 1. What are your conclusions?
Looking at the graphs, you wonder if forecasting with a time series is the best option. Since stocks run out when the product is on sale, maybe the orders would correlate better with the sales price of the product. To test your theory, you find historical sales price data as shown in Table 3.
4) Apply a linear regression forecasting model which correlates order quantities to price. Use the data from weeks 1 to 20 to find the linear equation. Then use this equation to calculate the forecasts starting from week 6. Use the same steps as for question 1 to analyse this model. What are your conclusions?
5) If you compare the TS and the mean absolute deviation (MAD) for the 3 methods tested using orders as the demand data, which forecasting method would you recommend?