代写Research Goal、代写c/c++程序语言、代做Java，Python设计

日期：2019-10-28 10:39

Assignment 2

Chocolate Demand Forecast

General Research Goal

In your role as a marketing analyst for Cadbury you are asked to forecast the immediate impact of

Cadbury’s pricing decisions on sales in the UK. Your research goal is to estimate a forecasting model

that is as accurate as possible based on the data that is available.

Data

In the dataset rawdata_assignment2.csv you will find information on Cadbury sales in 68 weeks in

the UK. Compared to the dataset we have used in the respective tutorial, the dataset has been

extended to contain more specific information on promotional activities of Cadbury and its

competitors. The variables contained in the dataset are:

• Sales1: for Cadbury (number of units sold in hundreds of thousands)

• Price1: $ Price for Cadbury

• Price2-4: $ Price for competitors Droste, Baronie, Delicata (in that order)

• Feature1: $ Spendings for feature-only for Cadbury in hundreds of thousands

• Feature2-4: $ Spendings for feature-only for Droste, Baronie, Delicata (in that order) in

hundreds of thousands

• Display1: $ Spendings for display-only for Cadbury in hundreds of thousands

• Display2-4: $ Spendings for display-only for Droste, Baronie, Delicata (in that order) in

hundreds of thousands

• Fand1: $ Spendings for feature and display for Cadbury in hundreds of thousands

• Fand3 and 4: $ Spendings for feature and display for Baronie and for Delicata (in that order)

in hundreds of thousands

Note that there is no variable on feature and display for Droste, as this brand did not use both

promotional techniques in the same week.

A display promotion concerns a built up display in a shop, such as shown in Exhibit 1 below. A

feature promotion concerns features of the products, such as 20% extra for free, a free raffle ticket,

etc. Exhibit 2 is an example of a feature promotion.

Exhibit 1: Example of a display promotion

Exhibit 2: Example of a feature promotion

Task 1 (5 points)

You want to estimate a linear regression but you know that price and promotional activities typically

have a nonlinear effect on sales so you decide to estimate elasticities. Thus, your first step is to

formulate the log-linear variables for all sales, price, and promotional variables. Please note that

taking logs requires positive values of sales, price and promotional variables; if a variable contains

zeros you will not be able to formulate log-linear values for them. The standard procedure to deal

with this problem is to identify the variables that contain a 0 and to simply add .01 to each

observation of the respective variables. You then have positive values and you can formulate a loglinear

transformation of the variables subsequently.

Please name all variables that required adding .01 to zeros.

Task 2 (30 points)

Please formulate a linear regression analysis with ln_sales as the dependent variable and the lnrepresentations

of the chosen independent variables. Please explain why you chose this model over

other alternatives. Please report your results, evaluate the model and discuss the effects.

Task 3 (35)

3a) Please use your model to predict weekly sales for a Cadbury price of $1.60. Please use your

model coefficients to do so as we did in the lecture and in the tutorial. Write down three steps: 1)

Write down your preferred model (from Task 2). 2) Assume that all other variables are zero (please

set all variables other than ln_price1 at zero) and add the price information to your model. 3) Predict

sales.

Tip: Remember that the model coefficients are elasticities! Also make sure that you provide

sufficient information so that we can evaluate your prediction so write down every step. In this task

it is not the final number that earns the points but the procedure you have followed.

3b) Please compare the price of $1.60 with the actual Cadbury prices in our dataset. Is it high or is it

low for Cadbury standards? How about your sales prediction? Please explain why or why not you

trust your results.

Task 4 (30)

Your Chief Marketing Office (CMO) sees great value in a sales forecasting model such as yours.

However, he prefers to weight all pros and cons before actually making model-based pricing

decision.

First, please explain to him what your regression model can do and what it cannot do. Second, point

out to him why you think the model offers a good start for further considerations about pricing

decisions. Third, given that you had to work with the data that is available so far, what could you do

to test your model? Fourth, highlight 5 additional variables you would consider useful should you be

expected to further improve the performance of the model in the future; please provide your

reasoning.

Assessment

- Please use the assignment form provided on Moodle

- Please use font size 12, double-spaced.

- Please do not submit more than 4 pages (including tables etc., excluding the cover page).

Penalties may apply for longer submissions.

- Please submit one hardcopy no later than Monday, 28 October, 2pm.

- Please submit it in the lecture.