Chapter 10 Machine Learning Algorithm Performance Metrics

Question 1

WHAT ARE THE TYPES OF CLASSIFICATION PERFORMANCE METRICS? P72

Answer 1

ˆ Classification Accuracy.
ˆ Logarithmic Loss.
ˆ Area Under ROC Curve.
ˆ Confusion Matrix.
ˆ Classification Report.

Question 2

HOW IS LOGARITHMIC LOSS (LOGLOSS) CALCULATED? P73

Answer 2

Logarithmic loss (or logloss) is a performance metric for evaluating the predictions of probabilities of membership to a given class, Predictions that are correct or incorrect are rewarded or punished proportionally to the confidence of the prediction.

Question 3

FOR WHAT KIND OF CLASSIFICATION PROBLEMS IS AREA UNDER ROC CURVE USED? MULTICLASS OR BINARY? P73

Answer 3

Area under ROC Curve (or AUC for short) is a performance metric for binary classification problems.

Question 4

WHAT DOES ROC-AUC REPRESENT? P73

Answer 4

The AUC represents a model’s ability to discriminate between positive and negative classes.

Question 5

WHAT PERFORMANCE METRIC A CONFUSION MATRIX SHOWS? P74

Answer 5

It’s a handy presentation of the accuracy of a model with 2 or more classes

Question 6

WHAT ARE THE 3 MOST COMMON METRICS FOR EVALUATING PREDICTIONS ON REGRESSION MACHINE LEARNING PROBLEMS? P76

Answer 6

ˆ Mean Absolute Error.
ˆ Mean Squared Error.
ˆ R2

Question 7

WHICH MODELS ASSUME GAUSSIAN DISTRIBUTION FOR NUMERICAL INPUTS? P80

Answer 7

Logistic Regression, Linear Discriminant Analysis, (Gaussian) Naïve Bayes
StackOverflowForMore
https://stats.stackexchange.com/questions/360657/list-of-machine-learning-classifiers-that-naturally-assume-data-in-normal-distri