Study sets, textbooks, questions
Upgrade to remove ads
Terms in this set (10)
Regularization (how does it affect to variance and bias)
Decreases variance and increases bias
Getting more training data (how does it affect to variance and bias)
No effect on bias and decreases variance.
Relu activation (what's the output for negative and positive value)
negative = 0, positive = positive
If the learning rate is too high, what can happen
the cost function may diverge away from an optimal solution
the cost function does not converge to an optimal solution
Suppose you use test data to tune a hyperparameter of your model. Which can happen
Using a cross validation set for tuning and a separate test set for measuring performance typically provides an unbiased, realistic measurement of performance.
Using test data to estimate performance accuracy will overestimate the accuracy of the model.
The model will not generalize well to unseen data because it overfits the data.
Does F1 score help imbalanced data or not?
Precision and recall are good metrics for evaluating models involving imbalanced classes
The accuracy score from an imbalanced data set provides no evidence that the model performs well?
training error is high & validation error is also high - what does it mean?
The model has high bias and underfits the training data
In case of high bias, would increasing more training data help the issue?
No (but yes for variance)
Sets with similar terms
Applied ML in python
Ch. 9 Big Data Analytics for Managing Risk
ITCS-3153 Exam 2
Other Quizlet sets
Earth Science 8, Chapter 14
Psych 101 CLEP practice test questions 2021
Chapter 15: Risk Mitigation
HA - Chapter 1