find_origin: Core functions to asses overall performance of LDA, NBC and...

In McCannLab/spatial_fingerprints: Research compendium

Description

Core functions to asses overall performance of LDA, NBC and MLP.

Usage

find_origin(
  method = c("lda", "nb", "ml"),
  df_dat,
  ndistr = 20,
  nsample = 10,
  noise = 0,
  col_ids = 3:19,
  toprob = TRUE
)

scr_lda(df_dat, ndistr = 20, nsample = 10, noise = 0, col_ids = 3:19)

scr_nb(df_dat, ndistr = 20, nsample = 5, noise = 0, col_ids = 3:19)

Arguments

method	statistical approach, either "lda", "nb" or "ml".
df_dat	bio-tracer data sets. Note that by default, it loads df_all() (see get_data_ready()).
ndistr	number of sample used to build baseline bio-tracer distributions.
nsample	number of samples.
noise	noise to be added.
col_ids	ids (name or integer) of columns to be used (select bio-tracers).
toprob	should the output be a probability.

Details

• "lda" : linear discriminant analysis,

• "nb" : naive bayesian,

• "ml" : machine learning,

• ndistr + nsample cannot exceed 30 with data in df_all.

Functions

• scr_lda: core function for LDA.

• scr_nb: core function for the Naive Bayesian (NB) classifier.

Examples

df_dat <- get_data_ready()
find_origin("nb", col_ids = 3:4, df_dat = df_dat)

McCannLab/spatial_fingerprints documentation built on March 13, 2021, 12:02 a.m.