predict_topics: predict_topics

In rosepeglershare/TagR: SHARE's TagR package that uses machine learning to identify a comment's relevant topics

Description

Trains an xgboost model for each topic and uses this to predict the probability that unlabelled comments belong to this topic or not.

Usage

predict_topics(
  unlabelled_raw,
  labelled_dtm,
  unlabelled_dtm,
  labels_matrix,
  text_vars,
  num_vars,
  topics,
  parameters = list(booster = "gbtree", objective = "binary:logistic", max_depth = 6,
    eta = 0.3, subsample = 1, colsample_bytree = 1, min_child_weight = 1),
  parameters_df = NULL,
  nrounds = 1000
)

Arguments

unlabelled_raw	Original unlabelled dataframe before any pre-processing.
labelled_dtm	Full labelled document-term matrix.
unlabelled_dtm	Unlabelled document-term matrix used for predictions.
labels_matrix	Labels matrix for labelled_dtm.
text_vars	List of text variables.
num_vars	List of numerical variables.
topics	List of topics.
parameters	Default list of parameters if user did not perform hyperparameter tuning.
parameters_df	A dataframe with columns representing parameters and rows representing an optimal parameter set for each topic.
nrounds	Number of rounds that the xgboost model should be trained for. Default: 1000

Value

A dataframe with the original comments, chosen attributes and probabilities that they belong to each topic.

rosepeglershare/TagR documentation built on Dec. 31, 2020, 3:12 a.m.