trellixvulnteam / 02_regression_gbhb Goto Github PK

Topic: Review of Regression via Sci-Kit Learn

Agenda:

• Tip: For datasets that display linear form when on a scatter plot, it may be beneficial to address with linear models. We usually fit a line to the data using least squares. The goal is to find the line that results in the minimum sum of squared residuals. This generates y = mx + b. Caution high variance may lead to overfitting.

1. Generalized Linear Models

• Ridge Regression -> Helps ensure the model does not overfit the training data. This is accomplished by introducing Bias to the model is fit to the data. This Bias should help reduce the variance. By providing less fit Ridge Regression may provide better long-term predictions.

2. Kernel Ridge Regression

3. Support Vector Machines

4. Stochastic Gradien Descent

5. Nearest Neighbors

6. Gaussian Processes

7. Decision Trees

8. Ensemble Methods

9. Multi-class and Multi-label algorithms

10. Isotonic Regression

11. Neural Network Models (supervised)