clustering

  ↳ stoclust

Contains functions providing basic clustering techniques motivated by stochastic analysis.

Functions

• fushing_mcassey

  fushing_mcassey(st_mat, max_visits=5, time_quantile_cutoff=0.95, index=None)
  Given a stochastic matrix describing the strength of the relationship between pairs of items, determines an aggregation of the items using the regulated random walk approach of Fushing and McAssey.
  

• hier_from_blocks

  hier_from_blocks(block_mats, scales=None, index=None)
  Given a parameterized ensemble of block matrices, each more coarse-grained than the last, constructs a corresponding Hierarchy object.
  

• meyer_wessell

  meyer_wessell(st_mat, min_times_same = 5, vector_clustering = None, index = None)
  Given a square column-stochastic matrix describing the strength of the relationship between pairs of items, determines an aggregation of the items using the dynamical approach of Meyer and Wessell..
  

• shi_malik

  shi_malik(st_mat, eig_thresh=0.95, tol=0, index=None)
  Given a square column-stochastic matrix describing the strength of the relationship between pairs of items, determines an aggregation of the items using the spectral approach of Shi and Malik.
  

• split_by_gaps

  split_by_gaps(vec, num_gaps = 1, index = None)
  Aggregates the indices of a vector based on gaps between index values. The number of gaps is specified, and the largest gaps in the sorted array are used to cluster values.
  

• split_by_quantiles

  split_by_quantiles(vec, quantiles=0.95, index = None)
  Cuts the vector at specific quantiles rather than rigid values. Assumes right-continuity of the cumulative distribution function.
  

• split_by_vals

  split_by_vals(vec, cuts=0, index = None, tol=0)
  Aggregates the indices of a vector based on specified values at which to cut the sorted array. Assumes the right-continuity of the cumulative distribution function.