ensemble_metrics: Compute ensemble metrics

In ThomasChln/opticskxi: OPTICS K-Xi Density-Based Clustering

Compute ensemble metrics

Description

Use models' rankings over several metrics to select best model. Several approaches can be taken to sum the models' rankings, and instead of summing the ranks of all models over all metrics, we prefer to rank only the top models for each metrics, and set 0 to all other. This behavior is controlled by the n_top parameter. In a second step, we sum the ranks and return only the top models, and this is controlled by the n_models parameter. The output is a list of the rankings matrix, for quality control purposes, and the selected models' parameters data frame, which is used by the ensemble_models function.

Usage

ensemble_metrics(
  n_top = 0,
  df_params,
  metrics = NULL,
  metrics_exclude = NULL,
  n_models = 10
)

Arguments

n_top	Threshold of number of models to rank
df_params	Output of opticskxi_pipeline
metrics	Names of metrics to use. Any of those computed by opticskxi_pipeline, e.g. 'sindex', 'ch', 'dunn', 'dunn2', 'widestgap', 'entropy' etc. NULL for all (8).
metrics_exclude	Names of metrics to exclude. Typically used with metrics = NULL. E.g. 'entropy'.
n_models	Number of best models to return

Value

List of metrics' rankings matrix and best models' parameters data frame.

ThomasChln/opticskxi documentation built on April 12, 2025, 5:43 a.m.