SigmaAnalytics & Tech
G2PythonAdvancedAbout 180 min to complete

Classification

Classification is the task of predicting a category: which species, spam or not, will churn or stay. You met one classifier already — logistic regression (Notebook 46).

Notebook

About 180 minutes to complete

Enrol to read the notebook and download it to run.

What this module covers

Classification is the task of predicting a category: which species, spam or not, will churn or stay. You met one classifier already — logistic regression (Notebook 46).

By the end of it

  • Use the uniform scikit-learn classifier API — fit / predict / predict_proba
  • Train and contrast the core classifiers: Logistic Regression, k-Nearest Neighbors, Decision Tree, Random Forest (plus SVM and Naive Bayes)
  • Evaluate with a confusion matrix, classification_report (precision/recall/F1), and ROC/AUC
  • Use cross-validation for a reliable performance estimate, and diagnose overfitting vs underfitting
  • Visualise decision boundaries to build intuition for each algorithm
  • Read feature importances and handle multiclass problems and class imbalance