market sale
So, what would you recommend that they order for the next season? That may not be the same as what they expect to sell. Is the manager’s practice of ordering 50% more than expected a good practice? What would you suggest as a better method? One way you can do this is to figure out how good was their method and yours in terms of MAD or Forecasting Accuracy. If your method is better, then there is no reason to order 50% more like they did. Often times businesses, especially smaller ones, either over order or under order because they don’t do a good enough analysis to dertermine. This is the idea of safety stocks.
In addition, even though we are discussing Multiple Regression Analysis this week, it does not mean you have to use that in the Hightower case. in other words, whether you use one independent variable or multiple independent variable is entirely up to whether it is approciate or not.
This is an example of using Regression. You can do that in Excel under Data Analysis again. Make sure you understand what is Y, the depend variable or the variable you want to forecast, and X, the independent variable or the variable’s information you use to forecast. Sometimes we mix up the two. Also, in this case, we are going to use only one independent variable, but sometimes you can have more. In Excel, if you have more, you just have to block them altogether. It is very simple, I still remember days in the past that we have to do this by hand, it required matrix and you don’t want to know!
First, how are we going to forecast. if you cut through all the noise of info, you can see that probably a very good indicator of holiday season sales is the test market sales. in other words, if test market sales is high, you would expect holiday sales to be high too. that is the concept of cause and effect, using an independent variable, test market sales, to forecast a dependent variable, holiday sales. sure there are probably some other factors/variables, but may be this is good enough. so regression analysis is for us to use historical data and find out the math relationship between our dependent var and independent var. that is, we are finding Y = a + bX, or Realized sales = a + b x Tested Sales. we take the given data from past years, go to Excel to Data, then Data Analysis, then Regression, Y is Realized sales, X is tested sales. you will find the relationship to be Realized Sales = 68 + 12.4 x Tested Sales.
So in order to make forecast for holiday Sales/Realized sales, you tested for bear, pigs, and racoons. so for bears, you got 10 for tested sales(or sales during test market), if you plug 10 into the equation, you should get 194, which should be the forecast for bears for the upcoming holiday season. do the same for pigs and racoons. that would be your forecast.
Anyway, for Hightower, we just want to establish a forecasting model that we can use to forecast holiday season sales in the future. There is a slight twist if you enjoy these things. That is, whatever you forecast, usually you want to order more than that, that is called the safety stock in business. She used 50% more. How about you? How are you going to determine how much more? Obviously if you order more and not use them, it is a waste and it costs you, if you order less, you won’t have enough and you miss profit opportunities. So how should you decide? This is a tough one and if you don’t want to try, just don’t. It is a mind twister that is useful in business but we can leave it to the math nuts!
But one thing you should do is know how they decide on the accuracy of your forecast model. That we do by comparing the forecast values with the actual values(those that we do have), that is the MAD. usually you would express that as a %. for example, you may have a forecasting accuracy of 85%, which means your forecasting error is 15%. Try that too with Hightower.
attachment
hightower_department_store-1.pdf
attachment
hightower_department_store-data_only.xls
attachment
hightower_departme
