Machine Learning Zoomcamp Update:Thursday, 30 September 2021
Date: 30 September 2021
Today, I completed the Sessions 4.5, 4.6 and 4.7.
Session 4.5 - ROC Curves
This session explained about ROC curve.
Key takeaways:
- ROC curve stands for Receiver Operating Characteristic curve.
-
False Positive Rate (FPR) = FP/(TN+FP)
True Positive Rate (TPR) = TP/(FN+TP) - TPR = Recall
Session 4.6 - ROC AUC
This session explained ROC AUC (Area Under Curve).
Key takeaways:
- ROC AUC >= 0.5, if it is not true then invert the predictions.
-
Ideal model ROC AUC = 1
Random model ROC AUC = 0.5 -
Probability that a randomly selected positive (true) example has higher prediction probability value than a randomly selected negative example = AUC
P(randomly selected positive example > randomly selected negative example) = AUC
Session 4.7 - Cross-Validation
This session gave an introduction to K-fold cross validation.
Estimated Time Taken: 1 hour 30 minutes