- Can Var% explained in the random forest replace R Square? And is Var% explained similar to the result of cross-validation?
Yu Shi
To always know where you are working, use the Working directory getwd().
1. Adding the dataset.
2. Encoding the target feature as a category variable.
3. Dividing the dataset into two parts: training and testing.
4. Scaling of features.
5. Adapting the Decision Tree to the Training Data.
6. Predict the results of the test set using Random Forest.
Random Forests may be used for regression problems in addition to classification. The nonlinear structure of a Random Forest might offer it an advantage over linear algorithms, making it an excellent choice.
Also, see https://stats.stackexchange.com/questions/376959/random-forest-var-explained-oob-output-different-from-test-dataset-results.
Yu Shi
In simple words, variation explained gives the goodness of the out-of-bag prediction explained targeted variance in the training dataset.
Again, R-squared is measured by 1-mse/var(y). Here R-squared also deals with explaining variation in the fitted model.
One condition can be brought on, if the error predictor is unbiased, then both variation explained and R-squared should have equal results.
I would say NO. Variation explained gives the goodness of the outside prediction explained targeted variance in the training dataset
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Biochemical Techniques
- Analytical Chemistry
- pH