How do you rate a model’s performance ? Depending on the context, this can lead to discussions about false positives/negatives, accuracy, recall, AUC, lift… How should one decide?

Data Science Interview Questions › Category: Data Science › How do you rate a model’s performance ? Depending on the context, this can lead to discussions about false positives/negatives, accuracy, recall, AUC, lift… How should one decide?
0 Vote Up Vote Down
Paras asked 6 years ago

Question taken from: Shape Science, blog on data science and algorithms

2 Answers
0 Vote Up Vote Down
MockInterview Staff answered 6 years ago
  • Type of task: regression? Classification?
  • Business goal?
  • What is the distribution of the target variable?
  • What metric do we optimize for?
  • Regression: RMSE (root mean squared error), MAE (mean absolute error), WMAE(weighted mean absolute error), RMSLE (root mean squared logarithmic error)…
  • Classification: recall, AUC, accuracy, misclassification error, Cohen’s Kappa…

SOURCE

0 Vote Up Vote Down
MockInterview Staff answered 6 years ago

If you want to visually remember the definitions of precision and recall, check out this quora answer by Jason T Widjaja: https://www.quora.com/What-is-the-best-way-to-understand-the-terms-precision-and-recall/answer/Jason-T-Widjaja?srid=uWIN
precision and recall
precision and recall

Your Answer

10 + 12 =