Icon

Challenge 26 - Modeling Churn Predictions IIII

Challenge 26 - Modeling Churn Predictions IIII
Challenge 26: Modeling Churn Predictions IIII Description: To wrap up our series of data classification challenges, consider again the following churning problem: a telecomcompany wants you to predict which customers are going to churn (that is, going to cancel their contracts) based on attributes oftheir accounts. The target class to be predicted in the test data is Churn (value 0 corresponds to customers that do not churn, and1 corresponds to those who do). You have already found a good model for the problem and have already engineered the trainingdata to increase the performance a bit. Now, your task is to communicate the results you found visually. Concretely, build adashboard that:1- shows performance for both classes (you can focus on any metrics here, e.g., precision and recall)2- ranks features based on how important they were for the model2- explains a few single predictions, especially false positives and false negatives, with our Local Explanation View component(read more about it here)https://www.knime.com/blog/xai-local-explanation-view-component DATA INPUT TRAIN MODELS & PREDICTION CHECK RESULTS H2O AutoML - StackedEnsembleRead Testing Data ~20%ReadTraining Data ~80%Node 3Acc: 96.1 %Cohen's kappa: 83.4 %K = 5Oversampleminority class(Churn = 1)Input:0 : Model as a Workflow Object1 : Data from Model Test PartitionOutput:0 : Global Feature ImportanceAutoML CSV Reader CSV Reader Workflow Executor Scorer (JavaScript) SMOTE Global FeatureImportance Challenge 26: Modeling Churn Predictions IIII Description: To wrap up our series of data classification challenges, consider again the following churning problem: a telecomcompany wants you to predict which customers are going to churn (that is, going to cancel their contracts) based on attributes oftheir accounts. The target class to be predicted in the test data is Churn (value 0 corresponds to customers that do not churn, and1 corresponds to those who do). You have already found a good model for the problem and have already engineered the trainingdata to increase the performance a bit. Now, your task is to communicate the results you found visually. Concretely, build adashboard that:1- shows performance for both classes (you can focus on any metrics here, e.g., precision and recall)2- ranks features based on how important they were for the model2- explains a few single predictions, especially false positives and false negatives, with our Local Explanation View component(read more about it here)https://www.knime.com/blog/xai-local-explanation-view-component DATA INPUT TRAIN MODELS & PREDICTION CHECK RESULTS H2O AutoML - StackedEnsembleRead Testing Data ~20%ReadTraining Data ~80%Node 3Acc: 96.1 %Cohen's kappa: 83.4 %K = 5Oversampleminority class(Churn = 1)Input:0 : Model as a Workflow Object1 : Data from Model Test PartitionOutput:0 : Global Feature ImportanceAutoML CSV Reader CSV Reader Workflow Executor Scorer (JavaScript) SMOTE Global FeatureImportance

Nodes

Extensions

Links

Download

To use this workflow in KNIME, download it from the below URL and open it in KNIME:

Download Workflow
Created by: USER
Created at: 2022-07-02
On NodePit since: 2022-08-31
Last update: 2024-04-26
Created with KNIME version: v4.5.0
Tags: justknimeit-26

Deploy, schedule, execute, and monitor your KNIME workflows locally, in the cloud or on-premises – with our brand new NodePit Runner.

Try NodePit Runner!