Scikit-learn Machine Learning Library

Scikit-learn (formerly scikits.learn) is a free and open-source machine learning library for Python that provides simple and efficient tools for data mining, data analysis, and predictive modeling. Originally created by David Cournapeau as a Google Summer of Code project in 2007, scikit-learn was further developed by Fabian Pedregosa, Gael Varoquaux, Alexandre Gramfort, and Vincent Michel at INRIA (the French Institute for Research in Computer Science). The first public release (v0.1) was in 2010. Built on top of NumPy, SciPy, and matplotlib, scikit-learn is the most widely used ML library for classical machine learning. Key features: supervised learning: classification (SVM, nearest neighbors, random forest, decision trees, naive Bayes, logistic regression, gradient boosting) and regression (linear regression, ridge, lasso, elastic net, SVR, random forest regressor). Unsupervised learning: clustering (K-Means, DBSCAN, hierarchical clustering, spectral clustering, mean-shift), dimensionality reduction (PCA, t-SNE, UMAP, factor analysis, NMF, ICA), and anomaly detection (Isolation Forest, One-Class SVM). Model selection: comprehensive tools for cross-validation (k-fold, stratified, leave-one-out), hyperparameter tuning (GridSearchCV, RandomizedSearchCV), and evaluation metrics (accuracy, precision, recall, F1, ROC-AUC, MSE, MAE). Preprocessing: feature scaling (StandardScaler, MinMaxScaler, RobustScaler), encoding (OneHotEncoder, LabelEncoder, OrdinalEncoder), imputation (SimpleImputer, KNNImputer), and feature selection (SelectKBest, RFE). Pipeline: chain multiple preprocessing and modeling steps into a single estimator for consistent training and deployment. Feature extraction: text (CountVectorizer, TfidfVectorizer) and image feature extraction. Built on NumPy arrays and SciPy sparse matrices for efficient computation. Partial fitting for out-of-core learning. Python. BSD-3-Clause.