Customer Segmentation and Marketing Strategy Advice from a Statistical Perspective

View this project on my Github: Link Here

First, we use hierarchical clustering to Identify the characteristics of different customer groups, divide customers into three groups - generous people, low-income people, Middle Group, and summarize the characteristics of the three groups of people, which will help the company improve marketing and promotional strategies. Then we use the method of multiple linear regression to conduct multiple regression analysis on consumption to study the relationship between each variable and consumption. We found that the logarithm of income has a significant impact on consumption, which is consistent with the classic psychological law; the larger the number of children, the lower the consumption; different commodities depend on different sales channels, online and offline sales should be more targeted.

Next, we use the machine learning model to predict the consumption, get a random forest model with better effect, and discuss the importance of each variable to the regression. Finally, we focus on how to improve the company’s promotional activities. By training the classification model to find the characteristics of people who are easy to accept promotional activities, and then make targeted recommendations to the company. In this way, we gain a full understanding of data, and also provide comprehensive suggestions for companies to understand their customers.