scikit-learn

name	scikit-learn
description	Machine learning library for classical supervised/unsupervised models, preprocessing, and model evaluation using NumPy/SciPy.
version	1.8.0
ecosystem	python
license	BSD-3-Clause
generated_with	gpt-5.2

Imports

import sklearn
from sklearn import datasets, metrics
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler
from sklearn.linear_model import LogisticRegression
from sklearn.neighbors import NearestNeighbors

Core Patterns

Check installation + environment details ✅ Current

import sklearn

if __name__ == "__main__":
    print("scikit-learn version:", sklearn.__version__)
    sklearn.show_versions()

• Use sklearn.show_versions() when debugging environment issues (BLAS/OpenMP, NumPy/SciPy versions, compiler info).
• sklearn.show_versions() output starts with "System:" in scikit-learn 1.8.0+.

Train/test split + fit/predict + metrics ✅ Current

from sklearn import datasets, metrics
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split

if __name__ == "__main__":
    X, y = datasets.load_breast_cancer(return_X_y=True)

    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=0.25, random_state=0, stratify=y
    )

    clf = LogisticRegression(max_iter=2000, solver="lbfgs")
    clf.fit(X_train, y_train)

    y_pred = clf.predict(X_test)
    print("accuracy:", metrics.accuracy_score(y_test, y_pred))
    print(metrics.classification_report(y_test, y_pred))

• Standard estimator workflow: fit() on train, predict() on test, evaluate with sklearn.metrics.

Pipeline with preprocessing + model ✅ Current

from sklearn import datasets, metrics
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler

if __name__ == "__main__":
    X, y = datasets.load_breast_cancer(return_X_y=True)
    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=0.25, random_state=0, stratify=y
    )

    pipe = Pipeline(
        steps=[
            ("scaler", StandardScaler()),
            ("clf", LogisticRegression(max_iter=2000)),
        ]
    )

    pipe.fit(X_train, y_train)
    y_pred = pipe.predict(X_test)
    print("accuracy:", metrics.accuracy_score(y_test, y_pred))

• Prefer Pipeline to prevent train/test leakage (fit transformers only on training data).

Hyperparameter search with GridSearchCV ✅ Current

from sklearn import datasets, metrics
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import GridSearchCV, train_test_split
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler

if __name__ == "__main__":
    X, y = datasets.load_breast_cancer(return_X_y=True)
    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=0.25, random_state=0, stratify=y
    )

    pipe = Pipeline(
        steps=[
            ("scaler", StandardScaler()),
            ("clf", LogisticRegression(max_iter=2000)),
        ]
    )

    param_grid = {
        "clf__C": [0.1, 1.0, 10.0],
        "clf__penalty": ["l2"],
        "clf__solver": ["lbfgs"],
    }

    search = GridSearchCV(
        estimator=pipe,
        param_grid=param_grid,
        scoring="accuracy",
        cv=5,
        n_jobs=-1,
    )
    search.fit(X_train, y_train)

    print("best params:", search.best_params_)
    y_pred = search.predict(X_test)
    print("test accuracy:", metrics.accuracy_score(y_test, y_pred))

• Use step__param names to tune pipeline components.
• Use n_jobs=-1 to parallelize where supported.

Nearest-neighbor index + query ✅ Current

import numpy as np
from sklearn.neighbors import NearestNeighbors

if __name__ == "__main__":
    rng = np.random.RandomState(0)
    X = rng.randn(10, 3)
    query = rng.randn(2, 3)

    nn = NearestNeighbors(n_neighbors=3, metric="euclidean")
    nn.fit(X)

    distances, indices = nn.kneighbors(query, return_distance=True)
    print("indices:\n", indices)
    print("distances:\n", distances)

• NearestNeighbors provides unsupervised neighbor search via fit() + kneighbors().

Configuration

• Randomness / reproducibility
  • Many APIs accept random_state=... (e.g., train_test_split, many estimators). Prefer setting it in tests and benchmarks.
  • Test reproducibility can be controlled via the environment variable SKLEARN_SEED (used by scikit-learn’s test suite).
• Installation verification
  • Prefer python -m pip ... to avoid mismatched interpreter vs pip.
  • Useful commands:
    • python -m pip show scikit-learn
    • python -m pip freeze
    • python -c "import sklearn; sklearn.show_versions()"
• Optional dependencies
  • Plotting helpers (functions named plot_... and classes ending with ...Display) require matplotlib.
• Running tests
  • Run from outside the source tree using pytest sklearn and ensure pytest meets the minimum required version (per scikit-learn’s developer docs).

Pitfalls

Wrong: Mixing OS package manager installs with pip installs (Linux)

# This is a shell sequence, not Python; shown here because it causes Python import/runtime issues.
# sudo apt-get install python3-sklearn
# pip install -U scikit-learn
import sklearn

print(sklearn.__version__)

Right: Use an isolated environment (venv) and verify with sklearn.show_versions()

# Shell sequence:
# python3 -m venv sklearn-env
# source sklearn-env/bin/activate
# python -m pip install -U scikit-learn
# python -c "import sklearn; sklearn.show_versions()"

import sklearn

if __name__ == "__main__":
    sklearn.show_versions()

Wrong: Using pip and python from different environments/interpreters

# Shell sequence that can target different interpreters:
# pip install -U scikit-learn
# python -c "import sklearn; sklearn.show_versions()"

import sklearn
print(sklearn.__version__)

Right: Always use python -m pip with the same interpreter you run

# Shell sequence:
# python -m pip install -U scikit-learn
# python -m pip show scikit-learn
# python -c "import sklearn; sklearn.show_versions()"

import sklearn

if __name__ == "__main__":
    sklearn.show_versions()

Wrong: Plotting API usage without Matplotlib installed

# If matplotlib is not installed, importing plotting helpers can fail at import time.
from sklearn.inspection import PartialDependenceDisplay  # requires matplotlib

print(PartialDependenceDisplay)

Right: Install matplotlib before using plot_* or *Display APIs

# Shell:
# python -m pip install -U scikit-learn matplotlib

from sklearn.inspection import PartialDependenceDisplay

if __name__ == "__main__":
    print("PartialDependenceDisplay import OK:", PartialDependenceDisplay)

Wrong: Data leakage by scaling before train/test split

from sklearn import datasets, metrics
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler

if __name__ == "__main__":
    X, y = datasets.load_breast_cancer(return_X_y=True)

    scaler = StandardScaler()
    X_scaled = scaler.fit_transform(X)  # leakage: uses all data statistics

    X_train, X_test, y_train, y_test = train_test_split(
        X_scaled, y, test_size=0.25, random_state=0, stratify=y
    )

    clf = LogisticRegression(max_iter=2000)
    clf.fit(X_train, y_train)
    print(metrics.accuracy_score(y_test, clf.predict(X_test)))

Right: Put preprocessing inside a Pipeline

from sklearn import datasets, metrics
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import StandardScaler

if __name__ == "__main__":
    X, y = datasets.load_breast_cancer(return_X_y=True)

    X_train, X_test, y_train, y_test = train_test_split(
        X, y, test_size=0.25, random_state=0, stratify=y
    )

    pipe = Pipeline(
        steps=[("scaler", StandardScaler()), ("clf", LogisticRegression(max_iter=2000))]
    )
    pipe.fit(X_train, y_train)
    print(metrics.accuracy_score(y_test, pipe.predict(X_test)))

References

• homepage
• source
• download
• tracker
• release notes

Migration from v1.7

• No breaking changes in Python version or core APIs between 1.7 and 1.8.0.
• If upgrading from 1.6 or older, remember: scikit-learn 1.7+ requires Python >= 3.10.
• For full migration notes and new features, see: https://scikit-learn.org/stable/whats_new/v1.8.0.html

# Example: Python version runtime guard (if upgrading from <=1.6)
import sys

if __name__ == "__main__":
    if sys.version_info < (3, 10):
        raise RuntimeError("scikit-learn >= 1.7 requires Python >= 3.10")
    print("Python version OK:", sys.version)

API Reference

• sklearn.version - The installed scikit-learn version (string, e.g., '1.8.0').
• sklearn.show_versions() - Print build/runtime dependency versions; use for debugging environment issues.
• *sklearn.clone(estimator, , safe=True) - Clone a scikit-learn estimator; returns a new, unfitted estimator with the same parameters.
• sklearn.get_config() - Get current global scikit-learn configuration as a dict.
• sklearn.set_config(assume_finite: bool = False, working_memory: int = 1024, print_changed_only: bool = True, display: str = 'text', pairwise_dist_chunk_size: int = 256) - Set global scikit-learn configuration.
• **sklearn.config_context(new_config) - Context manager for temporarily setting scikit-learn config.
• sklearn.datasets.load_breast_cancer(return_X_y=True) - Load a built-in dataset; return_X_y returns (X, y).
• sklearn.model_selection.train_test_split() - Split arrays into random train/test subsets; key params: test_size, random_state, stratify.
• sklearn.pipeline.Pipeline(steps=...) - Chain transformers and an estimator; prevents leakage; params: steps.
• sklearn.preprocessing.StandardScaler() - Standardize features by removing mean/scaling to unit variance.
• sklearn.linear_model.LogisticRegression() - Linear classifier; key params: C, penalty, solver, max_iter, random_state.
• sklearn.model_selection.GridSearchCV(estimator, param_grid, cv=..., scoring=...) - Exhaustive parameter search with CV; key params: param_grid, cv, n_jobs, scoring.
• sklearn.metrics.accuracy_score(y_true, y_pred) - Classification accuracy metric.
• sklearn.metrics.classification_report(y_true, y_pred) - Text summary of precision/recall/F1.
• sklearn.neighbors.NearestNeighbors(n_neighbors=..., metric=...) - Unsupervised neighbor search model.
• NearestNeighbors.fit(X) - Build neighbor index from training data.
• - Query nearest neighbors; returns .

Current Library State (from source analysis)

API Surface

{
  "library_category": "general",
  "apis": [
    {
      "name": "sklearn.clone",
      "type": "function",
      "signature": "clone(estimator, *, safe=True)",
      "signature_truncated": false,
      "return_type": "estimator",
      "module": "sklearn.base",
      "publicity_score": "high",
      "module_type": "public",
      "decorators": [],
      "deprecation": {
        "is_deprecated": false
      },
      "type_hints" 
         
           
           
           
        
         
           
           
           
        
      
    
    
       
       
       
       
       
       
       
       
       
       
         
      
       
    
    
       
       
       
       
       
       
       
       
       
       
         
      
       
         
           
           
           
        
         
           
           
           
        
         
           
           
           
        
         
           
           
           
        
         
           
           
           
        
      
    
    
       
       
       
       
       
       
       
       
       
       
         
      
       
    
    
       
       
       
       
       
       
       
       
       
       
         
      
       
    
    
       
       
       
       
       
       
       
       
       
       
         
      
       
         
           
           
           
        
      
    
    
       
       
       
       
       
       
       
       
       
       
         
      
    
    
       
       
       
       
       
       
       
       
       
       
         
      
       
    
    
    
    
  

Usage Patterns

[
  {
    "api": "DecisionTreeClassifier, ExtraTreeClassifier, RandomForestClassifier, ExtraTreesClassifier",
    "setup_code": [
      "import numpy as np",
      "from sklearn.datasets import make_classification",
      "from sklearn.tree import DecisionTreeClassifier, ExtraTreeClassifier",
      "from sklearn.ensemble import RandomForestClassifier, ExtraTreesClassifier"
    ],
    "usage_pattern": [
      "# Setup monotonicity constraints",
      "monotonic_cst = np.zeros(X.shape[1])",
      "monotonic_cst[0] = 1  # monotonic increasing on feature 0",
      "monotonic_cst[1] = -1 # monotonic decreasing on feature 1",
      "est = TreeClassifier(max_depth=None, monotonic_cst=monotonic_cst)",
      "# Optionally: est = TreeClassifier(max_depth=None, monotonic_cst=monotonic_cst, max_leaf_nodes=n_samples_train)",
      "if hasattr(est, 'random_state'): est.set_params(**{'random_state': global_random_seed})",
      "if hasattr(est, 'n_estimators'): est.set_params(**{'n_estimators': 5})",
      "if sparse_splitter: X_train = csc_container(X_train)",
      "est.fit(X_train, y_train)",
      "proba_test = est.predict_proba(X_test)"
    
     
      
      
      
      
      
      
    
     
      
      
      
    
     
  
  
     
     
      
      
      
      
    
     
      
      
      
      
      
      
      
      
      
      
      
      
      
    
     
      
      
    
     
      
      
    
     
  
  
     
     
      
      
    
     
      
      
      
      
      
      
    
     
      
    
     
      
      
    
     
  
  
     
     
      
    
     
      
      
      
      
    
     
      
    
     
      
      
    
     
  
  
     
     
      
      
    
     
      
      
      
      
      
      
    
     
      
    
     
      
      
    
     
  
  
     
     
      
    
     
      
      
      
      
    
     
      
    
     
      
      
    
     
  
  
     
     
      
    
     
      
      
      
      
      
      
    
     
      
      
    
     
      
      
    
     
  
  
     
     
      
      
    
     
      
      
      
      
      
      
      
      
      
      
    
     
      
      
    
     
      
    
     
  
  
     
     
      
      
    
     
      
      
      
      
      
      
      
    
     
      
      
    
     
      
    
     
  
  
     
     
      
      
    
     
      
      
      
      
      
      
      
    
     
      
    
     
      
    
     
  

Documentation & Changelog

{
  "documented_apis": [
    "sklearn.show_versions",
    "sklearnex.neighbors.NearestNeighbors"
  ],
  "conventions": [
    "Use virtual environments (venv or conda) for isolated and reproducible installations.",
    "Always activate your environment before running Python commands.",
    "Use pip or conda to install official releases of scikit-learn.",
    "Use functions starting with 'plot_' and classes ending with 'Display' for plotting capabilities.",
    "Check versions and installation with 'python -m pip show scikit-learn' and 'python -c \"import sklearn; sklearn.show_versions()\"'.",
    "Use binary wheels for NumPy and SciPy when installing with pip to avoid slow source builds.",
    "Follow the minimum dependency versions as listed in documentation for compatible operation.",
    "After installation, test your setup with 'pytest sklearn' (requires pytest).",
    "For Intel CPUs, consider using 'scikit-learn-intelex' for performance, but be aware solvers are not enabled by default."
  ],
  "pitfalls": [
    {
      "category": "Environment isolation",
      "wrong": "pip install -U scikit-learn  # in system Python or shared environment"
       
       
    
    
       
       
       
       
    
    
       
       
       
       
    
  
   
    
       
       
       
       
    
  
   

---