create_training_data: Feature Extraction

In aproudian2/rapt: Atom Probe Tomography Analysis

Feature Extraction

Description

function for extracting features for ML algorithm

Usage

create_training_data(
  i,
  pattern,
  rrl,
  nper = 1,
  params,
  pcp = 0.5,
  tol = 0.005,
  maxKr = 10,
  nKr = 200,
  maxGr = 5,
  nGr = 1000,
  maxGXGHr = 3,
  maxGXHGr = 8,
  nGXr = 1000,
  vside = 0.3,
  jitter = 0,
  feats = c("G_max_diff", "G_max_diff_r", "G_min_diff", "G_zero_diff_r", "F_min_diff",
    "F_min_diff_F", "Tm", "Rm", "Rdm", "Rddm", "Tdm", "GXGH_min_diff", "GXGH_95diff_r",
    "GXGH_FWHM"),
  total_time_start = as.numeric(Sys.time())
)

Arguments

i	an integer. This will be the iteration number. It determines the seed number in the clustersim function.
pattern	a 3D point pattern (object of type pp3) to use as the underlying point pattern (UPP) onto which generate clusters
rrl	output of average_relabelings. Used as expected values for summary functions
pcp	a numeric. Guest concentration
tol	a numeric. 'tol' or tolerance in clustersim
maxKr	a numeric. Maximum value to calculate K to in K3est
nKr	a numeric. Number of intervals at which to calculate K in K3est
maxGr	a numeric. Maximum value to calculate G to in G3est
nGr	a numeric. Number of intervals at which to calculate G in G3est
maxGXGHr	a numeric. Maximum value to calculate GXGH to in G3cross
maxGXHGr	a numeric. Maximum value to calculate GXHG to in G3cross
nGXr	a numeric. Number of intervals at which to calculate G in G3cross
vside	a numeric. voxel side length for F3est
jitter	a numeric. Scale of perturbations. 'radius' value in rjitter.pp3
feats	a vector. Names of all features to be kept
total_time_stat	a numeric. Leave as default to let it algorithm calculate iteration time

Details

Take an point pattern, simulate clustering, calculate summary functions on this new point pattern, extract features from summary functions.

aproudian2/rapt documentation built on Dec. 15, 2022, 4:24 a.m.