Predictive Model Accuracy Calculation Format

normalization
time_series
predictive_model

#1

Hello guys,
Which is the best form to calculate the predictive model accuracy, when data is normalized or in actual form?


#2

It depends upon which model is being used. Following are some of the measures of fit/accuracy normally used:

  1. Regression: R square and Standard Error of Regression (SER). People also use MAPE as a measure

  2. Logistic Regression: Hit rate, AUC, ROC, sensitivity, specifisity etc

  3. Decision Tree: Again Hit rate, AUC, ROC, sensitivity, specifisity etc

  4. Random Forest: OOB error along with Hit rate, AUC, ROC, sensitivity, specifisity etc