Setup & Basics
| Task | Code | Description |
|---|
| Import sklearn | import sklearn | Import library |
| Check version | sklearn.version | Display version |
| Import dataset utilities | from sklearn import datasets | Built-in datasets |
| Import model selection | from sklearn.model_selection import train_test_split | Split dataset |
| Import metrics | from sklearn import metrics | Evaluation metrics |
| Import preprocessing | from sklearn import preprocessing | Data preprocessing |
Loading Built-in Datasets
| Task | Code | Description |
|---|
| Load Iris dataset | from sklearn.datasets import load_iris <br> iris = load_iris() | Classic dataset |
| Load Digits dataset | from sklearn.datasets import load_digits | Handwritten digits |
| Load Wine dataset | from sklearn.datasets import load_wine | Wine classification |
| Load Breast Cancer dataset | from sklearn.datasets import load_breast_cancer | Cancer detection |
| Split data | X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42) | Train/test split |
Preprocessing
| Task | Code | Description |
|---|
| Standard scaling | from sklearn.preprocessing import StandardScaler <br> scaler = StandardScaler()<br>X_scaled = scaler.fit_transform(X) | Normalize features (mean=0, std=1) |
| Min-Max scaling | from sklearn.preprocessing import MinMaxScaler | Scale values between 0–1 |
| Label encoding | from sklearn.preprocessing import LabelEncoder<br>le = LabelEncoder() | Convert labels to integers |
| One-hot encoding | from sklearn.preprocessing import OneHotEncoder | Convert categorical to binary columns |
| Binarization | from sklearn.preprocessing import Binarizer(threshold=0.5) | Convert values above threshold to 1 |
| Polynomial features | from sklearn.preprocessing import PolynomialFeatures(degree=2) | Generate interaction terms |
Feature Selection
| Task | Code | Description |
|---|
| Variance threshold | from sklearn.feature_selection import VarianceThreshold | Remove low-variance features |
| Select K best | from sklearn.feature_selection import SelectKBest, f_classif | Keep top K features |
| Recursive feature elimination | from sklearn.feature_selection import RFE | Recursive feature selection |
| Feature importances | model.feature_importances_ | Get importance after fitting |
Supervised Learning Models
| Algorithm | Import | Description |
|---|
| Linear Regression | from sklearn.linear_model import LinearRegression | Predict continuous values |
| Logistic Regression | from sklearn.linear_model import LogisticRegression | Binary classification |
| Decision Tree | from sklearn.tree import DecisionTreeClassifier | Tree-based model |
| Random Forest | from sklearn.ensemble import RandomForestClassifier | Ensemble of trees |
| Support Vector Machine | from sklearn.svm import SVC | Max-margin classifier |
| Naive Bayes | from sklearn.naive_bayes import GaussianNB | Probabilistic model |
| K-Nearest Neighbors | from sklearn.neighbors import KNeighborsClassifier | Distance-based classifier |
| Gradient Boosting | from sklearn.ensemble import GradientBoostingClassifier | Boosted trees |
| AdaBoost | from sklearn.ensemble import AdaBoostClassifier | Adaptive boosting |
| Linear SVM | from sklearn.svm import LinearSVC | Fast linear classifier |
Unsupervised Learning Models
| Algorithm | Import | Description |
|---|
| K-Means Clustering | from sklearn.cluster import KMeans | Partition data into clusters |
| Hierarchical Clustering | from sklearn.cluster import AgglomerativeClustering | Tree-based clustering |
| DBSCAN | from sklearn.cluster import DBSCAN | Density-based clustering |
| Gaussian Mixture | from sklearn.mixture import GaussianMixture | Probabilistic clustering |
| PCA (Dim. Reduction) | from sklearn.decomposition import PCA | Reduce features |
| t-SNE | from sklearn.manifold import TSNE | Visualization for high-dim data |
Model Training & Prediction
| Task | Code | Description |
|---|
| Train model | model.fit(X_train, y_train) | Fit model to data |
| Make prediction | y_pred = model.predict(X_test) | Predict labels |
| Predict probability | model.predict_proba(X_test) | Class probabilities |
| Model coefficients | model.coef_ | Coefficients (linear models) |
| Model intercept | model.intercept_ | Intercept term |
Model Evaluation
| Metric | Code | Description |
|---|
| Accuracy | metrics.accuracy_score(y_test, y_pred) | Correct predictions |
| Precision | metrics.precision_score(y_test, y_pred) | Positive predictive value |
| Recall | metrics.recall_score(y_test, y_pred) | Sensitivity |
| F1 Score | metrics.f1_score(y_test, y_pred) | Harmonic mean |
| Confusion Matrix | metrics.confusion_matrix(y_test, y_pred) | Actual vs predicted |
| Classification Report | print(metrics.classification_report(y_test, y_pred)) | Summary of metrics |
| ROC Curve | metrics.roc_curve(y_test, y_score) | True/false positive trade-off |
| AUC Score | metrics.roc_auc_score(y_test, y_score) | Area under ROC |
Regression Metrics
| Metric | Code | Description |
|---|
| Mean Absolute Error | metrics.mean_absolute_error(y_true, y_pred) | Avg. absolute error |
| Mean Squared Error | metrics.mean_squared_error(y_true, y_pred) | Avg. squared error |
| RMSE | np.sqrt(metrics.mean_squared_error(y_true, y_pred)) | Root MSE |
| R² Score | metrics.r2_score(y_true, y_pred) | Coefficient of determination |
Model Selection & Validation
| Task | Code | Description |
|---|
| Train/Test split | train_test_split(X, y, test_size=0.2) | Random data split |
| K-Fold CV | from sklearn.model_selection import KFold | K-fold cross-validation |
| Cross-val score | from sklearn.model_selection import cross_val_score<br>cross_val_score(model, X, y, cv=5) | Evaluate model stability |
| Grid Search | from sklearn.model_selection import GridSearchCV | Hyperparameter tuning |
| Random Search | from sklearn.model_selection import RandomizedSearchCV | Faster hyperparameter tuning |
Pipeline
| Task | Code | Description |
|---|
| Import pipeline | from sklearn.pipeline import Pipeline | Chain steps together |
| Create pipeline | pipe = Pipeline([('scaler', StandardScaler()), ('model', LogisticRegression())]) | Combine preprocessing + model |
| Fit pipeline | pipe.fit(X_train, y_train) | Train pipeline |
| Predict pipeline | pipe.predict(X_test) | Predict via pipeline |
| Evaluate pipeline | pipe.score(X_test, y_test) | Score via pipeline |
Dimensionality Reduction
| Task | Code | Description |
|---|
| PCA | from sklearn.decomposition import PCA<br>pca = PCA(n_components=2) | Principal Component Analysis |
| Fit/Transform | X_pca = pca.fit_transform(X) | Apply PCA |
| Explained variance | pca.explained_variance_ratio_ | Variance explained by each PC |
| LDA | from sklearn.discriminant_analysis import LinearDiscriminantAnalysis | Linear discriminant analysis |
Visualization
| Task | Code | Description |
|---|
| Confusion matrix plot | from sklearn.metrics import ConfusionMatrixDisplay | Visual confusion matrix |
| ROC curve | from sklearn.metrics import RocCurveDisplay | Plot ROC curve |
| Precision-Recall curve | from sklearn.metrics import PrecisionRecallDisplay | Precision vs recall |
| Plot decision boundary | plt.contourf() | Visualize classification boundary |
Common Workflows
| Step | Code Example |
|---|
| 1️⃣ Load dataset | X, y = datasets.load_iris(return_X_y=True) |
| 2️⃣ Split data | X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2) |
| 3️⃣ Choose model | model = RandomForestClassifier() |
| 4️⃣ Train model | model.fit(X_train, y_train) |
| 5️⃣ Predict | y_pred = model.predict(X_test) |
| 6️⃣ Evaluate | metrics.accuracy_score(y_test, y_pred) |
Common Hyperparameters
| Model | Key Parameters |
|---|
| LogisticRegression | penalty, C, solver, max_iter |
| DecisionTreeClassifier | criterion, max_depth, min_samples_split |
| RandomForestClassifier | n_estimators, max_depth, random_state |
| SVC | kernel, C, gamma |
| KNeighborsClassifier | n_neighbors, weights, metric |
| GradientBoostingClassifier | n_estimators, learning_rate, max_depth |