Coverage for src/sensai/naive_bayes.py: 21%
53 statements
« prev ^ index » next coverage.py v7.6.1, created at 2024-08-13 22:17 +0000
« prev ^ index » next coverage.py v7.6.1, created at 2024-08-13 22:17 +0000
1import collections
2from math import log, exp
4import numpy as np
5import pandas as pd
7from .vector_model import VectorClassificationModel
10class CategoricalNaiveBayesVectorClassificationModel(VectorClassificationModel):
11 """
12 Naive Bayes with categorical features
13 """
14 def __init__(self, pseudo_count=0.1):
15 """
16 :param pseudo_count: the count to add to each empirical count in order to avoid overfitting
17 """
18 super().__init__()
19 self.prior = None
20 self.conditionals = None
21 self.pseudoCount = pseudo_count
23 def _fit_classifier(self, x: pd.DataFrame, y: pd.DataFrame):
24 self.prior = collections.defaultdict(lambda: 0)
25 self.conditionals = collections.defaultdict(lambda: [collections.defaultdict(lambda: 0) for _ in range(x.shape[1])])
26 increment = 1
27 for idxRow in range(x.shape[0]):
28 cls = y.iloc[idxRow, 0]
29 self.prior[cls] += increment
30 for idxFeature in range(x.shape[1]):
31 value = x.iloc[idxRow, idxFeature]
32 self.conditionals[cls][idxFeature][value] += increment
33 # get rid of defaultdicts, which are not picklable
34 self.prior = dict(self.prior)
35 self.conditionals = {k: [dict(d) for d in l] for k, l in self.conditionals.items()}
37 def _predict_class_probabilities(self, x: pd.DataFrame):
38 results = []
39 for _, features in x.iterrows():
40 class_probabilities = np.zeros(len(self._labels))
41 for i, cls in enumerate(self._labels):
42 lp = log(self._probability(self.prior, cls))
43 for idx_feature, value in enumerate(features):
44 lp += log(self._probability(self.conditionals[cls][idx_feature], value))
45 class_probabilities[i] = exp(lp)
46 class_probabilities /= np.sum(class_probabilities)
47 results.append(class_probabilities)
48 return pd.DataFrame(results, columns=self._labels)
50 def _probability(self, counts, value):
51 value_count = counts.get(value, 0.0)
52 total_count = sum(counts.values())
53 return (value_count + self.pseudoCount) / (total_count + self.pseudoCount)
55 def _predict(self, x: pd.DataFrame) -> pd.DataFrame:
56 results = []
57 for _, features in x.iterrows():
58 best_cls = None
59 best_lp = None
60 for cls in self.prior:
61 lp = log(self._probability(self.prior, cls))
62 for idxFeature, value in enumerate(features):
63 lp += log(self._probability(self.conditionals[cls][idxFeature], value))
64 if best_lp is None or lp > best_lp:
65 best_lp = lp
66 best_cls = cls
67 results.append(best_cls)
68 return pd.DataFrame(results, columns=self.get_predicted_variable_names())