npu.eval.kfold_evaluation_utils

Classes

KFoldEvaluationUtils()

class npu.eval.kfold_evaluation_utils.KFoldEvaluationUtils
Author:

Alberto M. Esmoris Pena

Util functions to handle the evaluation of k-folding procedures.

static generate_report_string(problem_name, metric_names, mus, sigmas, Q)

The string representation of the k-fold procedure report.

Parameters:

  • problem_name (str) – The name of the problem for which the k-folding was computed.

  • metric_names (list of str) – The names of the metrics monitored during the k-folding procedure.

  • mus – The vector of means such that each component represents the mean value of an evaluation metric used to assess the k-folding procedure.

Variables:

  • sigmas (np.ndarray) – The vector of standard deviations such that each component represents the standard deviation of an evaluation metric used to assess the k-folding procedure.

  • Q – The matrix of quantiles such that the component of each column vector represents the quantiles of an evaluation metric used to assess the k-folding procedure.

static do_kfold_plot(X, sigma, metric_names, **kwargs)

Do the plot of the k-fold procedure.

Parameters:

  • X (np.ndarray) – The matrix of quantitative evaluations. Each row must represent a fold and each column an evaluation metric. Thus, X[i][j] is the j-th evaluation metric on the i-th fold.

  • metric_names (list of str) – The names of the metrics monitored during the k-folding procedure.

  • kwargs – The key-word arguments governing the plot, e.g., the figsize or the suptitle.

Variables:

sigma (np.ndarray) – The vector of standard deviations such that each component represents the standard deviation of an evaluation metric used to assess the k-folding procedure.

Returns:

Nothing at all.