Practical Predictive Analytics

Running the prediction on the test sample

The eval_model function applies the rules of the prediction to the test data, and also produces the confusion matrix, for both absolute counts, and relative percentages, so there is no need to calculate the percentages manually, as we did above.

Applying the model results on the test sample shows the accuracy to be similar to the training sample, which somewhat validates the results.

The error rate is the percentages of wrong classifications, which is equivalent to 1 minus the accuracy rate:

prediction <- predict(model, test_data,type="class") #Evaluate prediction statistics  eval_model(prediction, test_data)