BA 875 Operations and Supply Chain Analytics
Operations and Supply Chain Analytics
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: THEend8_
BA 875
Operations and Supply Chain Analytics
Deliverable: Please submit your homework assignment as a single Python Notebook and clearly
state your answer to each component of a question. Make sure to save Notebook with output.
Forecasting Price Changes & Promotions
Dataset: jeans_data
You are given the same dataset of prices for a pair of jeans from a medium sized fashion retailer
in the southeast USA and their corresponding sales. The data is reported on a week-to-week
basis. In addition to price as an input, you are also given a binary indicator variable
(se_indicator) specifying whether a given week occurred during a special event (e.g., Christmas).
In the weekly historical data, there was an original retail price for the sales offered ($135) as well
as two promotional prices: 15% off ($101.25) and 25% off ($114.75). In a previous analysis (see
in-class example), the two 15% off and 25% off promotions that were already tested in stores
was examined to see if they had a statistically significant effect. This was done by specifying the
two promo prices as dummy variables, as follows:
= 0 + 1 ∗ 114 + 2 ∗ 101 + 3 ∗ e_indicator + ϵ
The above regression is great for retrospectively evaluating the two promotions. But it is not
prescriptive. We are interested in giving the retailer a sense of its estimated sales if it decided to
offer a new promo to predict sales for a week in the future, i.e., they want you to predict the sales
for a future week given you offer promo (P): “$25 Off” (not 25% off) promotion specifically
during a special event period. (Hint: after regression, apply the .predict() method).
As you are interested in predicting future sales from a new type of promotion, you no longer
need to examine the previous promotions separately (i.e., no longer need the two dummy
variables); you simply need to include “price” as the variable capturing price levels by week.
a. Model I: Price (in $) and special event indicator as an independent variable
• Calculate the predicted sales from promo (P) when directly including price and
special event indicator as independent variables into a regression model
b. Model I: Price (as a % discount off original price) and special event indicator as an
independent variable
• Calculate the predicted sales promo (P) when directly including price and special
event indicator as before, but this time replace price with a % discount variable, and
include discount as an independent variable into a regression model