We are presented with a sample of 4499 customers of a large Canadian bank that currently has a surplus of liability customers, i.e., depositors. The bank would like to grow their asset customer base, i.e., borrowers, through a targeted marketing campaign. Previous marketing efforts were able to garner a roughly 10% conversion rate, where a conversion was considered to be a sale of a personal loan product to an existing liability customer. From this campaign, which was held in Toronto, a data set was established and to be used to improve the efficiency of this year’s marketing campaign. This efficiency boost will stem from designing an algorithm to predict which clients are most likely to convert on the personal loan, and then focus marketing on this group. All sampled customers have a financial advisor/planner with this bank. The financial planners are mobile and can come to the customers if they wish so, however, listed in the data set are the branches that they most often serve. Financial advisors earn commission if they can advise customers to buy a personal loan product.
The data set includes the following variables: Age: Customer's age in completed years Experience: Number of years of professional experience Income: Annual income of the customer ($000) Branch Address: Address of customer’s home branch Family: Family size of the customer CCAvg: Avg. spending on credit cards per month ($000) Mortgage: Value of house mortgage if any ($000) Personal Loan: Did this customer accept the personal loan offered in the last campaign? Brokerage Account: Does the customer have a brokerage account with this bank? GIC: Does the customer have a Guaranteed Investment Certificate (GIC) account with this bank? Online: Does the customer use internet banking facilities? CreditCard: Does the customer use a credit card issued by this bank? Advisor Name: Customer’s financial advisor/planner with this bank Advisor Designation: Financial advisor’s designation
The project is divided into four parts:
Carefully inspect and clean the data. Use simple descriptive statistics and basic univariate plots. Report your findings and how you resolved any potential issues. Use the cleaned data set for next questions.
Ignoring the variables containing text, i.e., names, addresses and designations, explore the data with respect to the goal of the bank. Present your four most important results also using visualizations. What are your insights?
Who is or are the most successful financial advisor(s)? Discuss your results. Hint: How could “successful” be defined? There might be multiple possible definitions.
Partition the data into training (75%) and test (25%) sets. Use a number of different machine learning techniques and evaluate the algorithms. Show the classification matrix for the test data that results from using the best model. Clearly explain your results. How well did your model help you achieve the objective?