Marketing Analytics Capstone Project

Overview

This capstone project will give you an opportunity to apply what we have covered in the Foundations of Marketing Analytics specialization. By the end of this capstone project, you will have conducted exploratory data analysis, examined pairwise relationships among different variables, and developed and tested a predictive model to solve a marketing analytics problem. It is highly recommended that you complete all courses within the Foundations of Marketing Analytics specialization before starting the capstone course.

Syllabus

Marketing Analytics Project Description

This module will define the goals and activities for the marketing analytics capstone project.

Exploratory Analysis

In this module, we will begin to examine individual variables and their relationship to the status of the loan. Note, this module includes review items from previous courses in the specialization. This content is not required, but recommended as content to revisit.

Data Preparation and Model Building

While there are many ways to build a classification model, we will focus on using logistic regression, a common tool for marketing problems in which the dependent variable is binary. We will begin by choosing a single predictor variable and then determine which other variables need to be added to our analysis. In this module, we will focus on developing alternative models that all have a single predictor.

Model Validation and Comparison

In the previous module, we estimated a model linking home ownership to whether or not a loan is considered risky. In this module, we will begin by assessing the accuracy of this model relative to a naïve model. We will then use this spreadsheet as a means of assessing how well the model performs when different predictors are used.

Incorporating Multiple Predictor Variables

In this module, we will generalize the logistic regression tool that was developed to include multiple predictor variables. We will also consider an alternative means of evaluating the performance of the model.