Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
7 Cards in this Set
- Front
- Back
|
Training data |
Data that is used to fit the model |
|
|
Test data |
Data that is not used to fit the model |
|
|
Quality of fit |
MSE measures the quality of fit. With regression problems the quality of a model fit is determined by its MSE on the test data. For classification problems, the quality of the fit is measured by the proportion of cases for which classification is selected. |
|
|
MSE |
Training MSE is not a suitable indicator of model accuracy, more suitable indicator is " Test MSE". Where we use test data instead of training. |
|
|
MSE keypoints |
1.Training MSE < Test MSE at evey level of flexibility 2. Training MSE decreases as flexibility increases( however its not the case with test MSE) 3. Test MSE is a u-shaped => means accuracy is worse when f is either too inflexible or too flexible. |
|
|
Variance vs Bias |
• Method with higher flexibility has higher variance and low sq;bias • flexible estimators are sensitive to data input, hence have high variance Vice versa for inflexible. • when Variance and sq bias decrease at same rate. Test MSE is minimized. |
|
|
Accuracy with classification problems |
We use TEST ERROR RATE for "classification model error " |
