kdsplot#
- scikitplot.snsx.kdsplot(data=None, *, x=None, y=None, kind=None, n_deciles=10, round_digits=None, hue=None, weights=None, verbose=False, fill=False, line_kws=None, color=None, legend=True, ax=None, **kwargs)[source]#
Given labels y_true (0/1) and probabilities y_score arrays, compute a decile table.
The function sorts observations by descending score, assigns decile index (1..n_deciles) using pandas qcut on the rank/index to ensure near-equal bins, and computes standard decile-level stats:
[ "decile", "prob_min", "prob_max", "prob_avg", "cnt_resp", "cnt_resp_total", "cnt_resp_non", "cnt_resp_wiz", "cnt_resp_rndm", "rate_resp", "cum_resp", "cum_resp_pct", "cum_resp_total", "cum_resp_total_pct", "cum_resp_non", "cum_resp_non_pct", "cum_resp_wiz", "cum_resp_wiz_pct", "KS", "lift", ]
- Parameters:
- xstr, array-like
Ground truth (correct/actual) target values.
- ystr, array-like
Prediction probabilities for target class returned by a classifier/algorithm.
- n_decilesint, optional, default=10
The number of partitions for creating the table. Defaults to 10 for deciles.
- round_digitsint, optional, default=None
The decimal precision for the result. return df.round(round_digits)
- verbosebool, optional, default=False
If True, prints a legend for the abbreviations of decile table column names.
- Returns DataFrame indexed by decile (sorted ascending).
- Returns:
- pandas.DataFrame
The dataframe (decile-table) with the deciles and related information (decile-level metrics). If hue/facet semantics were used, the returned table will include extra columns for those keys (e.g., ‘hue’).
- Parameters:
- Return type:
