tune_lambda: Tune Lambda Weight and Number of Clusters

In activetext/activeR: a semi-supervised active learning algorithm for text classification.

Tune Lambda Weight and Number of Clusters

Description

Experiment function for tuning lambda.

Usage

tune_lambda(
  docs,
  docName = "text",
  index_name = "id",
  labelsName = "label",
  n_clusters = c(2),
  maxQuery = 1,
  active = FALSE,
  lambdas,
  seed,
  prop_init = 0.1,
  parallel = TRUE,
  k = 5
)

Arguments

docs	Dataframe for courpus.
docName	String; name of variable for document text.
index_name	String; name of variable for corpus index.
labelsName	String; name of variable for document labels.
n_clusters	vec; sequence of number of clusters to be tested.
maxQuery	int; max number of documents ot query per iteration.
lambdas	vec; sequence of lambda values to test.
seed	int; Seed value for randomization.
parallel	logical; Whether or not to parallelize the tuning process.
k	int; number for cross-validation.
prop_init;	float; Proportion of docs in each training set to directly label; the rest are used by the EM.

Value

Results from experiment. parallel select best # of clusters and lambda in terms of the mean F1

activetext/activeR documentation built on May 31, 2024, 10:21 a.m.