Permalink
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
50 lines (34 sloc) 2.9 KB

Define our Goal

Build a binary classification model to predict if a customer will make a term deposit.


Explain and Explore the Data

The data comes from direct marketing campaigns of a Portuguese banking institution. The marketing campaigns were based on phone calls. Often, more than one contact to the same customer was required, in order to access if a sign up for the product (term deposit) would be (yes) or (no).

Term deposits, also known as, a certificate of deposit, (CD), is a type of savings account that has a fixed interest rate and fixed date of withdrawal, known as the maturity date.

Typically there are no risks associated with these and the longer the term length, the more you will earn.

Data Dictionary


Insights

  • As successful outcomes of previous marketing campaigns increase by one unit, a customer is 1.9x as likely to sign up for a term deposit
  • As marketing activity increases in the months of March, October, December, and September by one unit respectively a customer is 1.3x, 1.17x, 1.15x, and 1.12x as likely to sign up for a term deposit
  • As the level of college education increases by one unit, a customer is 1.16x as likely to sign up for a term deposit
  • As the age group, GI Gen (90+), increases by one unit, a customer is 1.10x as likely to sign up for a term deposit

Recommendations

  • Tighten up data collection

    • Variables like “Job” are likely not an exhaustive representation of customers and have high potential for error if self-reported
    • Lots of variables contained a category entitled “Unknown” — is this synonymous with “Other” or “Null”
  • Target older, smarter, wealthier customers

    • These groups have shown us that they will convert
  • Run awareness campaigns to reach new customers in the older age groups (Baby Boomers, Silent Gen, GI Gen)

    • Currently they only make up 24% of our customer base Taking into account any filters that might point to wealthiness like household income or education status
  • Increase marketing activity in months with higher conversion rates

    • September, March, October, and December
    • Are there opportunities to run campaigns centered around tentpole holidays in these months? (e.g. Spring Forward with a new savings account or Kickstart your New Year Resolutions with a new savings account)
  • Evaluate telemarketing performance relative to other channels (e.g. direct mail, search, social, email, tv, etc.)? Does it compliment them?

    • May want to consider a divestment if ROI relative to other channels for goals (awareness, consideration, and conversions) is low across the board
  • Target customers who converted on past campaigns

    • This was the strongest predictor in our Logistic Regression model

    Check out my presentation here.