1. Learn
    2. /
  2. Courses
    3. /
  3. Machine Learning for Marketing Analytics in R
  • 1

    Modeling Customer Lifetime Value with Linear Regression

    Free

    How can you decide which customers are most valuable for your business? Learn how to model the customer lifetime value using linear regression.

  • 2

    Logistic Regression for Churn Prevention

    Predicting if a customer will leave your business, or churn, is important for targeting valuable customers and retaining those who are at risk. Learn how to model customer churn using logistic regression.

  • 3

    Modeling Time to Reorder with Survival Analysis

    Learn how to model the time to an event using survival analysis. This could be the time until next order or until a person churns.

  • 4

    Reducing Dimensionality with Principal Component Analysis

    CRM data can get very extensive. Each metric you collect could carry some interesting information about your customers. But handling a dataset with too many variables is difficult. Learn how to reduce the number of variables in your data using principal component analysis. Not only does this help to get a better understanding of your data. PCA also enables you to condense information to single indices and to solve multicollinearity problems in a regression analysis with many intercorrelated variables.

Initializing

Exercise

Looking at data

The dataset salesData is loaded in the workspace. It contains information on customers for the months one to three. Only the sales of month four are included. The following table gives a description of some of the variables whose meaning is less obvious.

Variable Description
id identification number of customer
mostFreqStore store person bought mostly from
mostFreqCat category person purchased mostly
nCats number of different categories
preferredBrand brand person purchased mostly
nBrands number of different brands

The packages readr, dplyr, corrplot, and ggplot2 have been installed and loaded.

Instructions

100 XP
  • Use the structure command str() in order to get an overview over the data.
  • Now visualize the correlation of the continuous explanatory variables for the past three months with the sales variable of this month. Use the functions cor() and corrplot() and the pipe operator. Note that the right variables have already been selected for you.
  • Additionally, make a boxplot displaying the distribution of the salesThisMon dependent on the levels of the categorical variable preferredBrand. The same has already been done for the categorical dependent variable mostFreqStore.