1#!/usr/bin/env python 

2# coding=utf-8 

3 

4 

5import tabulate 

6 

7from .measure import auc 

8 

9 

10def performance_table(data, fmt): 

11 """Tables result comparison in a given format 

12 

13 

14 Parameters 

15 ---------- 

16 

17 data : dict 

18 A dictionary in which keys are strings defining plot labels and values 

19 are dictionaries with two entries: 

20 

21 * ``df``: :py:class:`pandas.DataFrame` 

22 

23 A dataframe that is produced by our evaluator engine, indexed by 

24 integer "thresholds", containing the following columns: 

25 ``threshold``, ``tp``, ``fp``, ``tn``, ``fn``, ``mean_precision``, 

26 ``mode_precision``, ``lower_precision``, ``upper_precision``, 

27 ``mean_recall``, ``mode_recall``, ``lower_recall``, ``upper_recall``, 

28 ``mean_specificity``, ``mode_specificity``, ``lower_specificity``, 

29 ``upper_specificity``, ``mean_accuracy``, ``mode_accuracy``, 

30 ``lower_accuracy``, ``upper_accuracy``, ``mean_jaccard``, 

31 ``mode_jaccard``, ``lower_jaccard``, ``upper_jaccard``, 

32 ``mean_f1_score``, ``mode_f1_score``, ``lower_f1_score``, 

33 ``upper_f1_score``, ``frequentist_precision``, 

34 ``frequentist_recall``, ``frequentist_specificity``, 

35 ``frequentist_accuracy``, ``frequentist_jaccard``, 

36 ``frequentist_f1_score``. 

37 

38 * ``threshold``: :py:class:`list` 

39 

40 A threshold to graph with a dot for each set. Specific 

41 threshold values do not affect "second-annotator" dataframes. 

42 

43 

44 fmt : str 

45 One of the formats supported by tabulate. 

46 

47 

48 Returns 

49 ------- 

50 

51 table : str 

52 A table in a specific format 

53 

54 """ 

55 

56 headers = [ 

57 "Dataset", 

58 "T", 

59 "E(F1)", 

60 "CI(F1)", 

61 "AUC", 

62 "CI(AUC)", 

63 ] 

64 

65 table = [] 

66 for k, v in data.items(): 

67 entry = [ 

68 k, 

69 v["threshold"], 

70 ] 

71 

72 # statistics based on the "assigned" threshold (a priori, less biased) 

73 bins = len(v["df"]) 

74 index = int(round(bins * v["threshold"])) 

75 index = min(index, len(v["df"]) - 1) # avoids out of range indexing 

76 entry.append(v["df"].mean_f1_score[index]) 

77 entry.append( 

78 f"{v['df'].lower_f1_score[index]:.3f}-{v['df'].upper_f1_score[index]:.3f}" 

79 ) 

80 

81 # AUC PR curve 

82 entry.append( 

83 auc( 

84 v["df"]["mean_recall"].to_numpy(), 

85 v["df"]["mean_precision"].to_numpy(), 

86 ) 

87 ) 

88 lower_auc = auc( 

89 v["df"]["lower_recall"].to_numpy(), 

90 v["df"]["lower_precision"].to_numpy(), 

91 ) 

92 upper_auc = auc( 

93 v["df"]["upper_recall"].to_numpy(), 

94 v["df"]["upper_precision"].to_numpy(), 

95 ) 

96 entry.append(f"{lower_auc:.3f}-{upper_auc:.3f}") 

97 

98 table.append(entry) 

99 

100 return tabulate.tabulate( 

101 table, headers, tablefmt=fmt, floatfmt=".3f", stralign="right" 

102 )