I have a dataset of sales from a grocery chain in 17 different counties over a period of 77 weeks for different products. The time period is from week 27 in 2020, to week 52 in 2021. What I want to do is to analyze the effects of a campaign that lasted over a timeperiod of six weeks, from week 34-39 in 2021. I observe great seasonal effects when i study the dataset. I struggle with how to best control for the seasonality.
The dataset is based on weekly observations of sales data. To control for sesaonality we want to use monthly variation. I tried to control for these seasonal effects by creating montly dummies on the data from week 27 in 2020 to week 26 in 2021. I did this in different ways. Firstly I tried to run a linear regression of the whole dataset, where the dependent variable was my sales-numbers and my explainatory variables were my montly dummies. Secondly I tried to collapse(sum) my sales-numbers on a montly basis, and then I ran the command “regress” again with the same dependent and explainatory variables as before. My problem with this approach is that I am not able to exploit the data in the first 6 months of 2021. I also struggle with how to incorporate my coeffiecients for the monthly dummies into my panel-data-regression.
I feel as though i am a bit stuck, and I dont know what I should do. Do you have any suggestions for me on how to move forward, or what you think would be the best approach for removing the seasonal effects in my campaign-period? I really appreciate all help!
The dataset is based on weekly observations of sales data. To control for sesaonality we want to use monthly variation. I tried to control for these seasonal effects by creating montly dummies on the data from week 27 in 2020 to week 26 in 2021. I did this in different ways. Firstly I tried to run a linear regression of the whole dataset, where the dependent variable was my sales-numbers and my explainatory variables were my montly dummies. Secondly I tried to collapse(sum) my sales-numbers on a montly basis, and then I ran the command “regress” again with the same dependent and explainatory variables as before. My problem with this approach is that I am not able to exploit the data in the first 6 months of 2021. I also struggle with how to incorporate my coeffiecients for the monthly dummies into my panel-data-regression.
I feel as though i am a bit stuck, and I dont know what I should do. Do you have any suggestions for me on how to move forward, or what you think would be the best approach for removing the seasonal effects in my campaign-period? I really appreciate all help!
