8.2 Practical Machine Learning with R Churn Analysis (balanced with SMOTE)

0 ×

Churn Analysis (unbalanced with SMOTE)

This workflow repeats the Churn Analysis from the textbook Practical Machine Learning with R (https:\\ai.lange-analytics.com). We are using unbalanced data for a churn analysis. "Unbalanced" means that one class (customers who did not churn) contains significantly more observations than the other class (customers who churned). Check the Value Count Node (Churn Count Before SMOTE). Here, we use SMOTE to balance the data (see Value Count Node (Churn Count After SMOTE). Check the Sensitivity and Specificity in the Scorer to see how much better the model performs compared to the "unbalanced data" model.


URL: ai.lange-analytics.com https://ai.lange-analytics.com
URL: Here you can open the R analysis with SMOTE in RStudio https://ai.lange-analytics.com/exc/?file=09-LogRegrExerc200.Rmd
URL: A workflow for the "unbalanced data" model is in this space https://hub.knime.com/-/spaces/-/~XZgXHCZkJ0rAsvsy/current-state/
URL: Contact the author https://ai.lange-analytics.com/EmailForw.html

Nodes

• Logistic Regression Predictor2 ×
• Scorer2 ×
• Value Counter2 ×
• CSV Reader1 ×
• Column Filter1 ×
• Show all 11 nodes

Extensions

• FeatureKNIME Base nodes

Links

• No links available