Source code for revscoring.scoring.statistics.classification.counts

import logging
from collections import OrderedDict

from tabulate import tabulate

from ... import util
from ...model_info import ModelInfo

logger = logging.getLogger(__name__)


class Counts(ModelInfo):

    def __init__(self, labels, score_labels, prediction_key):
        super().__init__()
        self['n'] = len(score_labels)

        self['labels'] = OrderedDict((l, 0) for l in labels)
        self['predictions'] = OrderedDict(
            (actual, OrderedDict((predicted, 0) for predicted in labels))
            for actual in labels)

        for score, label in score_labels:
            predicted = score[prediction_key]
            self['labels'][label] += 1
            self['predictions'][label][predicted] += 1

    def format_str(self, path_tree, **kwargs):
        if len(path_tree) > 0:
            logger.warn("Ignoring path_tree={0!r} while formatting counts."
                        .format(path_tree))
        formatted = "counts (n={0}):\n".format(self['n'])
        table_data = [
            [repr(label), self['labels'][label], '-->'] +
            [count for pred, count in pred_counts.items()]
            for label, pred_counts in self['predictions'].items()]
        table_str = tabulate(
            table_data, headers=['label', 'n', ''] +
                                ["~{0}".format(l) for l in self['labels']])
        formatted += util.tab_it_in(table_str)
        return formatted

    def format_json(self, path_tree, **kwargs):
        return util.dict_lookup(self._data, path_tree)


class MultilabelCounts(ModelInfo):

    def __init__(self, labels, score_labels, prediction_key):
        super().__init__()
        self['n'] = len(score_labels)

        self['labels'] = OrderedDict((l, 0) for l in labels)
        self['predictions'] = OrderedDict(
            (target_label,
             OrderedDict((actual,
                          OrderedDict((predicted, 0)
                                      for predicted in [True, False]))
                         for actual in [True, False]))
            for target_label in labels)

        for score, label in score_labels:
            label_set = set(label)
            predicted_set = set(score[prediction_key])
            for label in label_set:
                self['labels'][label] += 1
            for l in labels:
                self['predictions'][l][l in label_set][l in predicted_set] += 1

    def format_str(self, path_tree, **kwargs):
        if len(path_tree) > 0:
            logger.warn("Ignoring path_tree={0!r} while formatting counts."
                        .format(path_tree))
        formatted = "counts (n={0}):\n".format(self['n'])
        table_data = [
            [repr(label), self['labels'][label], '-->'] +
            [lstats[a][p] for a in [True, False] for p in [True, False]]
            for label, lstats in self['predictions'].items()]
        table_str = tabulate(
            table_data, headers=['label', 'n', '', 'TP', 'FP', 'FN', 'TN'])
        formatted += util.tab_it_in(table_str, tabs=2)
        return formatted

    def format_json(self, path_tree, **kwargs):
        return util.dict_lookup(self._data, path_tree)