check_accuracy: Function to check reconstruction accuracy

In EandrewJones/mmBPFA: Multi-Modal Beta-Process Factor Analysis

Description

check_accuracy ingests the cleaned results from process_results without standard errors and returns the accuracy in reconstructing the observed data. If true_vals from generate_sim_data is supplied, the function will also return the RMSE and MAE of posterior predictions for the latent values. A confusion matrix, precision, recall, and Kappa score are also available via confusion = TRUE for data with fixed marginal distributions.

Usage

check_accuracy(dat, results, true_vals = NULL, mode, confusion = T)

Arguments

dat	Data matrix.
results	processed results from process_results. Must not include standard errors.
true_vals	Optional list of true latent values from generate_sim_data.
mode	Margin type. Must be one of "fixed", "multi" or "mixed".
confusion	Boolean indicating whether to return a confusion matrix. Only works for "fixed" margin data.

Value

List of observed accuracy and accuracy metrics (rmse and mae). If confusion is TRUE, also returns confusion matrix results.

EandrewJones/mmBPFA documentation built on Feb. 14, 2021, 11:17 p.m.