Python Tutorial: Classification models

Want to learn more? Take the full course at https://learn.datacamp.com/courses/model-validation-in-python at your own pace. More than a video, you'll learn hands-on coding & quickly apply skills to your daily work. --- Welcome back - In this lesson, we switch from regression to classification models. This lesson focuses on reviewing classification models— or models built for when the response variable is categorical. Predicting a newborn's hair color, the winner of a basketball game, or the genre of the next song to come on the radio station are all examples of categorical responses - and we can build a classification model for each of them. When looking at classification models during this course, we will primarily use the Tic-Tac-Toe end-game dataset. This dataset contains the complete set of possible configurations at the end of a game of Tic-Tac-Toe. Each of the first nine columns represents one of the nine squares of a Tic-Tac-Toe board. A "b" means the square is blank, an "X" represents player one, and an "O" is for player two. The final column indicates if the first player won (labeled positive) or not (labeled negative). The tic_tac_toe dataset is ideal for model validation because we have the complete set of outcomes. We can include as much, or as little, of this data in our models as we want. This allows us to really test how well the model is performing on unseen data. And if you just got an urge to play Tic-Tac-Toe, Google will play against you as long as you would like! Several methods are shared across all scikit-learn models, but some are unique to the specific type of model. Before, we used the .predict() method to predict the value of new observations. scikit-learn's classifier, RandomForestClassifier() also has the method .predict(). This time, the new class of the observations is returned. We can also view how many observations were assigned to each class by turning the array of predictions into a pandas Series, and then using the method .v