Key Points

A Brief Introduction to Machine Learning

---

• There are many types of machine learning.
• We will focus on some methods that work well with tabular data.

Linear and Logistic Regression

---

• Classical linear and logistic regression models can be thought of as examples of regression and classification models in machine learning.
• Testing sets can be used to measure the performance of a model.

Decision Trees

---

• Training data can give us a decision tree model.
• Decision trees can be used for supervised learning, but they are not very robust.

Random Forests

---

• Random forests can make predictions of a categorical or quantitative variable.
• Random forests, with their default settings, work reasonably well.

Gradient Boosted Trees

---

• Gradient boosted trees can be used for the same types of problems that random forests can solve.
• The learning rate can affect the performance of a machine learning algorithm.

Cross Validation and Tuning

---

• Parameter tuning can improve the fit of an XGBoost model.
• Cross validation allows us to tune parameters using the training set only, saving the testing set for final model evaluation.