Evaluation and Optimization of Machine Learning Algorithms in Personalized Marketing

With the development of society and the continuous updating of economic models, the importance of marketing for business operations is becoming increasingly high. However, due to the traditional marketing model adopting a popular strategy, personalized recommendations are insufficient. In recent years, machine learning algorithms have been widely applied and developed rapidly in various industries. Therefore, this article studied the effectiveness evaluation and optimization of machine learning algorithms in personalized marketing, and discussed the business recommendation marketing of banks as the research point. The study was conducted using the dataset provided by the Kaggle data competition website. Firstly, customer information was analyzed and divided into 12 explanatory variables and 1 response variable. Next, the Apache Nifi tool was used to collect information, ensuring the security of customer information. Then, the Informatica Data Quality tool was used to clean the information data and organize it into features that can be utilized by the algorithm. The chi square test was used to test the correlation of features, determine the value of
$a$
as 0.05, and select the appropriate feature component to input into the model. Finally, the GBDT (Gradient Boosting Decision Trees) algorithm was used to analyze features, determine the actual needs of customers, and provide personalized recommendations to customers. The research results showed that the model had a 94% precision in judging the personal wishes of customers, and had strong anti-interference ability and good robustness. This study can accurately infer customers' intentions based on customer information, providing a new approach for exploring personalized marketing.