Distance-based Analysis of DAta-manifolds in python (DADApy)

DADApy is a python package for distance-based analysis of data-manifolds.

The code can be found on GitHub at https://github.com/sissa-data-science/DADApy.

Table of Contents

• Installation
  • Installing the requirements
  • Installing the package
• Implemented Algorithms
  • Intrinsic dimension estimation
  • Density estimation
  • Density based clustering
  • Metric space comparison
  • Feature weighting
• Typical usage of the package
• Modules (API reference)
  • The base module
  • The id_estimation module
  • The density_estimation module
  • The clustering module
  • The id_discrete module
  • The metric_comparisons module
  • The feature_weighting module
  • The data module
  • The utils module
• Tutorial: Density Based Clustering
  • Plot of the estimated densities
  • Plot of the DP clustering assignation
  • Plot of the ADP clustering assignation
  • Dendrogram obtention and interpretation
  • 2D multidimensional scaling obtention and interpretation
  • Matrix obtention and interpretation
• Tutorial: Information Imbalance
  • 3D Gaussian with small variance along \(z\)
  • 4D isotropic Gaussian
  • 10D isotropic Gaussian dataset
  • Sinusoidal function
• Tutorial: Intrinsic dimension
  • Two NN estimator
  • Analysis of the ID of noisy datasets: twoNN vs gride
  • Scale analysis when the intrinsic dimension is high
• Tutorial: Using DADApy to analyse a molecular dynamics trajectory of a small peptide
  • Analysing the trajectory using selected dihedral angles
  • Analysing the trajectory using all heavy atom distances
  • Check equivalence of the two data descriptions
• Tutorial: Intrinsic dimension estimation for discrete metrics
• Tutorial: Differentiable Information Imbalance
  • Comparison with the standard Information Imbalance
  • Optimization on a 10D anisotropic Gaussian dataset
  • Backward greedy elimination
  • L1 regularization (lasso)
• Hands-on tutorial at SISSA
• Citing DADApy

Indices and tables

• Index

• Module Index

• Search Page