Summary - Classification

In this group you learned classification, accuracy, and confusion matrix.

Summary: classification

In this group:

Classification predicts labels; logistic regression produces probabilities with sigmoid.
In a binary problem predict_proba[:, 1] is the positive probability; 0.5 is only the default threshold.
With many classes each probability row sums to 1; OvR extends binary models.
Accuracy misleads on imbalanced data; DummyClassifier gives a baseline.
The confusion matrix separates TP, TN, FP and FN errors.

Exercise Classification report

Split Iris with test_size=0.3, random_state=0, stratify=y and train a LogisticRegression(max_iter=1000). Print the test accuracy with round(..., 3), then the confusion matrix.

Show example solution

from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.metrics import accuracy_score, confusion_matrix

X, y = load_iris(return_X_y=True)

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=0, stratify=y)
m = LogisticRegression(max_iter=1000).fit(X_train, y_train)
p = m.predict(X_test)
print(round(accuracy_score(y_test, p), 3))
print(confusion_matrix(y_test, p))

Test yourself

What should the first comparison be to show that a model is worthwhile?
- A baseline that predicts the most frequent class
- Training accuracy
- Another dataset
- No comparison is needed
Show answer

A baseline that predicts the most frequent class — The model must clearly beat a baseline that learns nothing.

Compute the confusion matrix.

cm = ____(y_test, y_pred)

Show answer

cm = confusion_matrix(y_test, y_pred)

Match the tools.

LogisticRegression

Probabilistic classifier
DummyClassifier

Baseline
accuracy_score

Share correct
confusion_matrix

Error breakdown

Next lesson: Train-test split

We Are Enhancing Your Experience

Cookie Preferences

Strictly Necessary Cookies

Analytical Cookies