—Individual Research Project—
1. Background Information
Bank X has previously sold products to its customers via rather broad sales campaigns where
customers were cold-called by a call-center. In retrospective the bank found that those campaigns
rarely paid-off because the costs for such an untargeted campaign turned out to be immense given
the small number of customers for which the campaign triggered a purchase.
The bank therefore decided to adopt a more targeted approach in its future campaigns. You are one
of the bank’s marketing analysts and you are asked for your help in that matter.
In particular, the bank plans to run a campaign to sell term deposits to its existing customers. A term
deposit is a fixed-term investment which involves a money deposit at a financial institution. When
buying a term deposit the money is usually locked in for the period for which the deposit is
advertised. Funds can only be withdrawn after the term ends. The advantage of term deposits over
other investment products is that the return on investment is very safe and the interest rate offered
is usually more attractive than the one offered by alternative products.
The bank’s management further informs you that the idea is to run rather small campaigns and to
target a limited number of customers. The bank is reconsidering the call-centre approach through
which customers will be contacted in the next campaign so we do not have information about the
expected costs for the next campaign. We only know that up to 1000 customers may be contacted.
We also know that a positive response is approximately worth $200 for the bank. Fully convinced by
your skills as a marketing analyst, the bank’s management is also keen to learn about the maximum
costs of contacting a customer in order to avoid making a loss by contacting a customer based on
the current data set and under the above assumptions (only 1000 customers contacted, $200 for a
positive response).
2. Data, Sample and Variables
Since the bank has run previous campaigns you are provided a rich dataset that contains information
about bank clients, details about the contacts made in the last campaign and other previous
campaign attributes. The dataset is called “bank campaign”. The dataset contains the following
variables in particular.
2.1. Bank client information
age: (numeric)
job: type of job (categorical: 'admin.','bluecollar','entrepreneur','housemaid','management','retired','selfemployed','services','student','technician','unemployed','unknown')
marital: marital status (categorical: 'divorced','married','single','unknown'; note: 'divorced' means
divorced or widowed)
education: (categorical: primary, secondary, tertiary and unknown)
default: has credit in default? (categorical: 'no','yes','unknown')
housing: has housing loan? (categorical: 'no','yes','unknown')
loan: has personal loan? (categorical: 'no','yes','unknown')
balance: Balance of the individual.
2.2. Details about the last contact made in the most current campaign
contact: contact communication type (categorical: 'cellular','telephone')
month: last contact month of year (categorical: 'jan', 'feb', 'mar', ..., 'nov', 'dec')
duration: last contact duration, in seconds (numeric). Important note: Obviously the duration of the
last call is not known before the call is performed. Consider very well, therefore, if this is a variable
that is useful to identify the customers that shall be contacted in a marketing campaign and to finetune
the campaign to them.
2.3. Campaign attributes
campaign: number of contacts performed during this campaign and for this client (numeric, includes
last contact)
pdays: number of days that passed by after the client was last contacted from a previous campaign
(numeric; -1 means client was not previously contacted)
previous: number of contacts performed before this campaign and for this client (numeric)
poutcome: outcome of the previous marketing campaign (categorical:
'failure','nonexistent','success')
2.4. Output variable
Respons - has the client subscribed a term deposit? (binary: 'yes','no')
3. Your Task
Your research goal is to develop a model that can be used to finetune the campaign and to target
prospects. You also want to derive the maximum costs for a customer contact to avoid making a
loss.
Since your work shall lay the foundation for future work on customer prospect identification by
yourself and by your marketing analytics colleagues at the bank, you are asked to provide excellent
documentation of your work in the form of a research report.
Please specify and estimate a logistic regression analysis that serves to fulfils the research goal.
Produce a report that contains the substance for other marketing analysts to critically evaluate your
research approach but is still appealing to a managerial audience.
It is recommended to check the suggested structure for the report on Moodle; you are free,
however, to choose whatever structure you think works best.
The report should not be longer than 10 pages including all graphs and tables but excluding
references (font size 12, double-spaced). Please use the form provided on Moodle as a cover page
for your report. Submit a hardcopy.
For the deadlines please see the updated timeline on Moodle.