Hello! At the moment I’m studying the “Customer Analytics in Python” course. I was about to do the Brand Choice Models homework, the final lecture of the Modeling Brand Choice section and I suddenly got confused about something: I think there might be a mistake in the example. To calculate the Own Price Brand Choice Elasticity and the Cross Price Brand Choice Elasticity for different segments, I think we have to take different coefficients, regarding what we have obtained for that segment. Following the course example: For the average customers, we obtain the next coefficients Data Frame: https://drive.google.com/file/d/18OwEB-jZIpdTx-B8-_jH4ydp1B3eW0gp/view Since in the example we are interested in Brand 5, we select the appropriate coefficient and we name it beta5 (-1.09). https://drive.google.com/file/d/1IU20BNGmult3CVcbXbouzZQdd2WX80s6/view And we use it to calculate the Price Brand Choice Elasticity for average customers. But… When in the example are obtained the coefficients for the Segment ‘Well-Off’, the Data Frame displays the following values: https://drive.google.com/file/d/1d6Mf_hLD_3TDbcEbl2D_O3twndO4QNFk/view and we obtain the value of -0.44, but we don’t use it and I think that it is the value we must use. Instead, we still use the old value beta5 = -1.09 as we can see in the figure below, extracted directly from the code in the lecture: https://drive.google.com/file/d/1mH_Pq-l2aC5uTHY2C2GbYED8yBthF1bV/view Am I wrong in my reasoning? Is there something I am not understanding? Is a mistake in the lecture? Thank you in advance for your answer, and I wish you a Very Merry Christmas and a Happy New Year. Jesus Rangel.

Hi Jesus, I understand your confusion. You could skip calculating the table altogether and still obtain the results. The reason why we originally included the table is to be able to track some dependencies between brand coefficients and price, based on the specific segments. However, they are not included in the computation of the Price Elasticity curve as the coefficients there are based on the brand only. I might update the Notebooks and leave this part of the code out, as I think you're right - it's more confusing than helpful in the context of the course. Hope this clarifies the issue! Best, 365 Eli

Hi Jesus, thanks for reaching out and for your question! The idea here is that though we're examining cross brand 4, our own brand remains brand 5, this is why we still use the same coefficient. When examining the price elasticity of a brand (or indeed with most of marketing analysis) we're interested in our own brand and that's the perspective we take a look at. In our case the beta coefficient is the coefficient for the Brand variable, and not for the segment variable. Thus, it does not change across segments, it would change for a different brand. Hope this helps! Best, The 365 Team

Hello Eli, thanks for your answer. I understand that I have to take the coefficient for Brand 5, and I agree with that. But I still don't understand 1 fact and I have a doubt about 1 idea: Analyzing directly the exercise provided in the lecture, we have: - Fact: We are selecting Training Data from Segment “Well-Off”, then and naturally we obtain the coefficients for Segment “Well-Off”, BUT we don’t use them, instead we still use the old beta5 = -1.09, the coefficient for the whole Brand and no particular Segment filtered, then Why do we calculate the coefficients for Segment “Well-Off”? - Idea: Maybe I´m wrong, but I understand that we are calculating the coefficients based on Training Data exclusively from Segment “Well-Off” to obtain the coefficients for all Brands and, of course, exclusively for Segment “Well-Off”, as you do in the lecture, and so, to use the proper coefficient for Brand 5 and Segment “Well-Off” as you don’t do in the lecture. I think the selected coefficient must change due to the change in Training Data to fit the model, being selected the coefficient for Brand 5, but the one that fits the data, I mean, for the Segment “Well-Off” = -0.44. The picture at the link below was taken directly from the example provided on your lecture, nothing was changed except I added one last line to see the actual value of beta5: https://drive.google.com/file/d/1x0lKFLZjtZGOXjz9zZbqm9bGjjZXoktq/view

Hi Jesus, the reason that we use the coefficient for brand 5 is that the coefficient is determined by the brand and not the customer segment. Hope this helps! Best, 365 Eli

Hi Eli, I´m sorry, but that was your answer since the first time I submitted the first question, but the second question is different and remains without answer: I agree using the coefficient for brand 5, but the one you calculated later in you example, I mean the one with the data from segment ´Well-off´: If you will use the coefficient for brand 5 with value -1.09, then and please tell me ¿Why do you calculate the coefficient for brand 5 and SEGMENT ´Well-off´ (with value -0.44) and NOT going to use it? I mean, ¿Why do you calculated the coefficients on that dataframe? You calculate them in your example but you NEVER use them. Please check the picture below: Expanding my clarification: Maybe I´m not explaining myself well. I decided to take a screenshot right from the lecture, so you can see it is taken right from the lecture and what are the bunch of coefficients that were calculated and not used anywhere, any of them (Coefficients for Segment ´Well-Off´). I want to know why those coefficients were calculated and not used. What´s its purpose in the lecture?: