Icon

01_​Analyze_​Training_​a_​Churn_​Predictor

Analyze Data by Training a Decision Tree Classifier for Churn Prediction
Pre-processing / Data Preparation Read Data from differentfiles - Contract Data - Calls Data Color Churn 0 -> Blue 1 -> Red Conversions - Convert the column "Churn"and "Area Code" to String Partitioning 80% training 20% testing Train Model Decision tree Save to PMML Evaluation Apply and score Model How to train Classification Model? Step 1: Drag the "Decision Tree Learner" node anddouble click to open the dialog Step 2: Select the "Class column" as "Churn" and"Quality measure" as "Gini Index"Step 3: RIght Click on the node and select "Executeand Open View" to train the model and to get a view ofthe Decision Tree How to evaluate Classification Model? Step 1: Drag the Decision Tree Predictor node. Step 2: Connect the output of "Decision TreeLearner" node to Port 0 and Test Dataset to Port 1.Execute the nodeStep 3: Connect the Predictor Output to "ROCCurve" and "Scorer" node to evaluate the model onvarious evaluation measures 80%vs. 20%Port 0: Train setPort 1: Test setcolorby churnclass = churnarea codeand churn ->StringAuCCalls datacontract dataperformancescoringapply decision tree Partitioning Color Manager DecisionTree Learner Number To String ROC Curve Excel Reader PMML Writer CSV Reader Scorer Decision TreePredictor Joiner Pre-processing / Data Preparation Read Data from differentfiles - Contract Data - Calls Data Color Churn 0 -> Blue 1 -> Red Conversions - Convert the column "Churn"and "Area Code" to String Partitioning 80% training 20% testing Train Model Decision tree Save to PMML Evaluation Apply and score Model How to train Classification Model? Step 1: Drag the "Decision Tree Learner" node anddouble click to open the dialog Step 2: Select the "Class column" as "Churn" and"Quality measure" as "Gini Index"Step 3: RIght Click on the node and select "Executeand Open View" to train the model and to get a view ofthe Decision Tree How to evaluate Classification Model? Step 1: Drag the Decision Tree Predictor node. Step 2: Connect the output of "Decision TreeLearner" node to Port 0 and Test Dataset to Port 1.Execute the nodeStep 3: Connect the Predictor Output to "ROCCurve" and "Scorer" node to evaluate the model onvarious evaluation measures 80%vs. 20%Port 0: Train setPort 1: Test setcolorby churnclass = churnarea codeand churn ->StringAuCCalls datacontract dataperformancescoringapply decision tree Partitioning Color Manager DecisionTree Learner Number To String ROC Curve Excel Reader PMML Writer CSV Reader Scorer Decision TreePredictor Joiner

Nodes

Extensions

Links