/// <summary>
/// Reduce X to its most important features.
/// </summary>
/// <param name="selector"></param>
/// <param name="x">[n_samples, n_features] The input samples.</param>
/// <param name="threshold"> The threshold value to use for feature selection. Features whose
/// importance is greater or equal are kept while the others are
/// discarded.</param>
/// <returns>[n_samples, n_selected_features]
/// The input samples with only the selected features</returns>
public static Matrix <double> Transform(
this ILearntSelector selector,
double[,] x,
ThresholdChoice threshold)
{
return(Transform(selector, x.ToDenseMatrix(), threshold));
}
/// <summary>
/// Reduce X to its most important features.
/// </summary>
/// <param name="selector"></param>
/// <param name="x">[n_samples, n_features] The input samples.</param>
/// <param name="threshold"> The threshold value to use for feature selection. Features whose
/// importance is greater or equal are kept while the others are
/// discarded.</param>
/// <returns>[n_samples, n_selected_features]
/// The input samples with only the selected features</returns>
public static Matrix <double> Transform(
this ILearntSelector selector,
Matrix <double> x,
ThresholdChoice threshold)
{
if (threshold == null)
{
throw new ArgumentNullException("threshold");
}
// Retrieve importance vector
var importances = selector.FeatureImportances();
if (importances == null)
{
throw new InvalidOperationException(
"Importance weights not computed. Please set" +
" the compute_importances parameter before " +
"fit.");
}
/*
* elif hasattr(self, "coef_"):
* if self.coef_.ndim == 1:
* importances = np.abs(self.coef_)
*
* else:
* importances = np.sum(np.abs(self.coef_), axis=0)
*
* else:
* raise ValueError("Missing `feature_importances_` or `coef_`"
* " attribute, did you forget to set the "
* "estimator's parameter to compute it?")
*/
if (importances.Count != x.ColumnCount)
{
throw new InvalidOperationException("X has different number of features than" +
" during model fitting.");
}
/*
* // Retrieve threshold
* if threshold is None:
* if hasattr(self, "penalty") and self.penalty == "l1":
# the natural default threshold is 0 when l1 penalty was used
# threshold = getattr(self, "threshold", 1e-5)
# else:
# threshold = getattr(self, "threshold", "mean")
*/
double thresholdvalue = threshold.GetValue(importances);
var mask =
importances.GetIndexedEnumerator().Where(v => v.Item2 >= thresholdvalue).Select(v => v.Item1).ToArray();
if (mask.Any())
{
return(x.ColumnsAt(mask));
}
else
{
throw new InvalidOperationException("Invalid threshold: all features are discarded.");
}
}
