{
  "name": "Scikit-Learn Feature Scalers and Transformers Comparison Matrix",
  "description": "A comparative technical matrix evaluating mathematical formulations, bounded output ranges, outlier robustness, sparsity handling, and empirical research applications across core scikit-learn feature scaling estimators, compiled from official scikit-learn API specifications and user guide documentation.",
  "page": "https://casrai.org/guides/sklearn-minmaxscaler",
  "publisher": "CASRAI",
  "license": "https://creativecommons.org/licenses/by/4.0/",
  "dateModified": "2026-11-14T19:04:08Z",
  "temporalCoverage": "2011/2026",
  "spatialCoverage": "Global",
  "columns": [
    {
      "name": "Scaler Name",
      "unit": "",
      "description": "Class identifier in the sklearn.preprocessing module"
    },
    {
      "name": "Mathematical Transformation",
      "unit": "",
      "description": "Formulation applied to feature vector X"
    },
    {
      "name": "Output Range",
      "unit": "",
      "description": "Default numerical bounds of scaled features"
    },
    {
      "name": "Outlier Robustness",
      "unit": "",
      "description": "Susceptibility of scaling parameters to extreme values"
    },
    {
      "name": "Sparse Matrix Support",
      "unit": "",
      "description": "Native compatibility with scipy.sparse matrices without dense expansion"
    },
    {
      "name": "Preserves Zeros",
      "unit": "",
      "description": "Whether the transformation maintains exact zero entries"
    },
    {
      "name": "Primary Research Application",
      "unit": "",
      "description": "Optimal algorithmic or domain modeling context"
    }
  ],
  "rows": [
    {
      "Scaler Name": "MinMaxScaler",
      "Mathematical Transformation": "(X - X_min) / (X_max - X_min) * (max - min) + min",
      "Output Range": "[0, 1] (configurable)",
      "Outlier Robustness": "Low (extrema heavily distorted by outliers)",
      "Sparse Matrix Support": "Conditional (only when min=0 and preserves zero)",
      "Preserves Zeros": "No (shifts values unless min is zero)",
      "Primary Research Application": "Neural networks, bounded pixel intensities, and KNN distance equalization"
    },
    {
      "Scaler Name": "StandardScaler",
      "Mathematical Transformation": "(X - mean) / standard_deviation",
      "Output Range": "Unbounded (~[-3, 3] for normal distributions)",
      "Outlier Robustness": "Low (mean and sample variance skewed by outliers)",
      "Sparse Matrix Support": "Yes (with with_mean=False)",
      "Preserves Zeros": "Only when with_mean=False",
      "Primary Research Application": "Principal component analysis, linear regression, and logistic regression"
    },
    {
      "Scaler Name": "RobustScaler",
      "Mathematical Transformation": "(X - median) / IQR (75th - 25th percentile)",
      "Output Range": "Unbounded",
      "Outlier Robustness": "High (median and interquartile range resist outliers)",
      "Sparse Matrix Support": "Yes (with with_centering=False)",
      "Preserves Zeros": "Only when with_centering=False",
      "Primary Research Application": "Biomedical assays, sensor telemetry, and financial data with heavy tails"
    },
    {
      "Scaler Name": "MaxAbsScaler",
      "Mathematical Transformation": "X / max(|X|)",
      "Output Range": "[-1, 1]",
      "Outlier Robustness": "Low (maximum absolute value sensitive to outliers)",
      "Sparse Matrix Support": "Yes (native compatibility)",
      "Preserves Zeros": "Yes (scales without centering offset)",
      "Primary Research Application": "Sparse text classification, word frequency matrices, and high-dimensional sparse arrays"
    },
    {
      "Scaler Name": "Normalizer",
      "Mathematical Transformation": "X / ||X||_p (per-sample vector norm)",
      "Output Range": "Unit norm (L1, L2, or max norm = 1.0)",
      "Outlier Robustness": "Moderate (scales across sample features rather than population columns)",
      "Sparse Matrix Support": "Yes (csr_matrix supported for L1 and L2)",
      "Preserves Zeros": "Yes (zero features remain zero)",
      "Primary Research Application": "Text mining, cosine distance classification, and directional vector comparisons"
    },
    {
      "Scaler Name": "QuantileTransformer",
      "Mathematical Transformation": "Empirical cumulative distribution mapped to uniform or Gaussian",
      "Output Range": "[0, 1] (uniform) or Unbounded (normal)",
      "Outlier Robustness": "Very High (non-linear mapping compresses extreme tails)",
      "Sparse Matrix Support": "Yes (with ignore_implicit_zeros=True)",
      "Preserves Zeros": "Only with ignore_implicit_zeros=True",
      "Primary Research Application": "Non-parametric feature alignment, multimodal data, and skewed biomarker distributions"
    }
  ],
  "sources": [
    {
      "title": "Scikit-Learn API Reference: sklearn.preprocessing",
      "url": "https://scikit-learn.org/stable/modules/classes.html#module-sklearn.preprocessing",
      "accessed": "2026-11-14"
    },
    {
      "title": "Scikit-Learn User Guide: Preprocessing Data (Section 6.3)",
      "url": "https://scikit-learn.org/stable/modules/preprocessing.html",
      "accessed": "2026-11-14"
    },
    {
      "title": "Scikit-Learn Example Gallery: Compare the Effect of Different Scalers on Data with Outliers",
      "url": "https://scikit-learn.org/stable/auto_examples/preprocessing/plot_all_scaling.html",
      "accessed": "2026-11-14"
    },
    {
      "title": "Pedregosa et al. (2011), Scikit-learn: Machine Learning in Python",
      "url": "https://jmlr.org/papers/v12/pedregosa11a.html",
      "accessed": "2026-11-14"
    }
  ]
}