Q1easyWhat is the primary purpose of KFold Cross-Validation? cv_test.pyATo increase the number of features.BTo ensure the model's performance isn't just a result of a lucky train/test split.CTo make the training process 5 times faster.DTo delete 80% of the training data.Check AnswerReveal Answer & ExplanationDiscuss on Board
Q2easyIn a regression task, what does Mean Squared Error (MSE) measure?AThe standard deviation of the input data.BThe average absolute difference between prediction and truth.CThe total number of correct predictions.DThe average of the squares of the errors (differences between prediction and truth). from sklearn.model_selection import KFold kf = KFold(n_splits=5)Check AnswerReveal Answer & ExplanationDiscuss on Board
Q3easyWhat is the F1-Score?AThe average of accuracy and error.BThe square root of the MSE.CThe sum of TP and TN.DThe harmonic mean of Precision and Recall.Check AnswerReveal Answer & ExplanationDiscuss on Board
Q4mediumIn a confusion matrix, if you have 80 True Positives, 10 False Positives, and 10 False Negatives, what is the Precision?A0.90B0.80C0.50D0.88Check AnswerReveal Answer & ExplanationDiscuss on Board
Q5easyIf the AUC-ROC score of a model is 0.5, what does it mean?AThe model is good but needs tuning.BThe model is worse than random guessing.CThe model is perfect.DThe model's performance is no better than random guessing.Check AnswerReveal Answer & ExplanationDiscuss on Board
Q6mediumA cancer detection model must minimize False Negatives (missing a sick patient) because it is dangerous. What metric should the team optimize for?AAccuracyBSpecificityCPrecisionDRecall (Sensitivity)Check AnswerReveal Answer & ExplanationDiscuss on Board
Q7easyWhat is Precision?ATP / Total positives.BFP / (FP + FN)CTP / (TP + FP)DTN / Total negatives.Check AnswerReveal Answer & ExplanationDiscuss on Board
Q8easyWhat is Specificity?ATN / (TN + FP)BThe number of true positives.CThe speed of the model.DTP / (TP + FN)Check AnswerReveal Answer & ExplanationDiscuss on Board
Q9easyA model has a high Bias. What does the learning curve look like?AHigh training error and high test error (plateauing at a high level).BLow training error, low test error.CRandom fluctuations.DLow training error, high test error.Check AnswerReveal Answer & ExplanationDiscuss on Board
Q10mediumYou are optimizing a model. Your training error is 0.1%, but your testing error is 25%. What is this?AData leakageBUnderfittingCOverfittingDBalanced modelCheck AnswerReveal Answer & ExplanationDiscuss on Board