[sklearn] useful sklearn stuff #sklearn

clf_and_plot_cm.py

from sklearn.model_selection import train_test_split
from sklearn import metrics # for evaluation
from sklearn.ensemble import RandomForestClassifier

# initiate a classifier and train on some data
rf = RandomForestClassifier(n_jobs=-1)
rf.fit(x_train, y_train)

# predict
y_predict = rf.predict(x_train)

# make confusion matrix using sklearn
cm = metrics.confusion_matrix(y_train, y_predict)

# note: normalize not implemented, might be useful to show % instead of raw numbers.
def plot_cm(cm, labels=labels, normalize=False):
    """takes in a confusion matrix as well as the label names"""
    df = pd.DataFrame(cm, columns=labels, index=labels)
    fig, ax = plt.subplots(figsize=(10,6))
    ax.set_title("Confusion Matrix")
    sns.heatmap(df, annot=True, fmt="d", annot_kws={"size": 8}, ax=ax, cmap="YlGnBu")
    ax.set_xlabel("Predicted label"); ax.set_ylabel("True label")
    plt.show()