The classification challenge

Apprendimento supervisionato con scikit-learn

George Boorman

Core Curriculum Manager, DataCamp

Classifying labels of unseen data

Build a model
Model learns from the labeled data we pass to it
Pass unlabeled data to the model as input
Model predicts the labels of the unseen data

Labeled data = training data

k-Nearest Neighbors

Predict the label of a data point by
- Looking at the k closest labeled data points
- Taking a majority vote

k-Nearest Neighbors

scatter plot with observations in blue and red, and a new observation in black

k-Nearest Neighbors

radius around the three observations nearest to the black dot

k-Nearest Neighbors

radius around the five observations nearest to the black dot

KNN Intuition

scatterplot of total evening charge versus total day charge, where observations are colored in blue if they have churned and red if they have not churned

KNN Intuition

churn scatterplot with a decision boundary splitting observations by whether KNN predicts they will churn or not

Using scikit-learn to fit a classifier

from sklearn.neighbors import KNeighborsClassifier

X = churn_df[["total_day_charge", "total_eve_charge"]].values
y = churn_df["churn"].values

print(X.shape, y.shape)

(3333, 2), (3333,)

knn = KNeighborsClassifier(n_neighbors=15)

knn.fit(X, y)

Predicting on unlabeled data

X_new = np.array([[56.8, 17.5],
                  [24.4, 24.1],
                  [50.1, 10.9]])

print(X_new.shape)

(3, 2)

predictions = knn.predict(X_new)

print('Predictions: {}'.format(predictions))

Predictions: [1 0 0]

Let's practice!

Apprendimento supervisionato con scikit-learn