Mercurial > repos > bgruening > sklearn_data_preprocess

comparison association_rules.py @ 37:1bef885255e0 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/sklearn commit ea12f973df4b97a2691d9e4ce6bf6fae59d57717"
author bgruening
date Sat, 01 May 2021 01:41:14 +0000
parents
children a16f33c6ca64
comparison
equal deleted inserted replaced
36:b75cae00f980 37:1bef885255e0
1 import argparse
2 import json
3 import warnings
4
5 import pandas as pd
6 from mlxtend.frequent_patterns import association_rules, fpgrowth
7 from mlxtend.preprocessing import TransactionEncoder
8
9
10 def main(inputs, infile, outfile, min_support=0.5, min_confidence=0.5, min_lift=1.0, min_conviction=1.0, max_length=None):
11 """
12 Parameter
13 ---------
14 input : str
15 File path to galaxy tool parameter
16
17 infile : str
18 File paths of input vector
19
20 outfile : str
21 File path to output matrix
22
23 min_support: float
24 Minimum support
25
26 min_confidence: float
27 Minimum confidence
28
29 min_lift: float
30 Minimum lift
31
32 min_conviction: float
33 Minimum conviction
34
35 max_length: int
36 Maximum length
37
38 """
39 warnings.simplefilter('ignore')
40
41 with open(inputs, 'r') as param_handler:
42 params = json.load(param_handler)
43
44 input_header = params['header0']
45 header = 'infer' if input_header else None
46
47 with open(infile) as fp:
48 lines = fp.read().splitlines()
49
50 if header is not None:
51 lines = lines[1:]
52
53 dataset = []
54 for line in lines:
55 line_items = line.split("\t")
56 dataset.append(line_items)
57
58 # TransactionEncoder learns the unique labels in the dataset and transforms the
59 # input dataset (a Python list of lists) into a one-hot encoded NumPy boolean array
60 te = TransactionEncoder()
61 te_ary = te.fit_transform(dataset)
62
63 # Turn the encoded NumPy array into a DataFrame
64 df = pd.DataFrame(te_ary, columns=te.columns_)
65
66 # Extract frequent itemsets for association rule mining
67 # use_colnames: Use DataFrames' column names in the returned DataFrame instead of column indices
68 frequent_itemsets = fpgrowth(df, min_support=min_support, use_colnames=True, max_len=max_length)
69
70 # Get association rules, with confidence larger than min_confidence
71 rules = association_rules(frequent_itemsets, metric="confidence", min_threshold=min_confidence)
72
73 # Filter association rules, keeping rules with lift and conviction larger than min_liftand and min_conviction
74 rules = rules[(rules['lift'] >= min_lift) & (rules['conviction'] >= min_conviction)]
75
76 # Convert columns from frozenset to list (more readable)
77 rules['antecedents'] = rules['antecedents'].apply(list)
78 rules['consequents'] = rules['consequents'].apply(list)
79
80 # The next 3 steps are intended to fix the order of the association
81 # rules generated, so tests that rely on diff'ing a desired output
82 # with an expected output can pass
83
84 # 1) Sort entry in every row/column for columns 'antecedents' and 'consequents'
85 rules['antecedents'] = rules['antecedents'].apply(lambda row: sorted(row))
86 rules['consequents'] = rules['consequents'].apply(lambda row: sorted(row))
87
88 # 2) Create two temporary string columns to sort on
89 rules['ant_str'] = rules['antecedents'].apply(lambda row: " ".join(row))
90 rules['con_str'] = rules['consequents'].apply(lambda row: " ".join(row))
91
92 # 3) Sort results so they are re-producable
93 rules.sort_values(by=['ant_str', 'con_str'], inplace=True)
94 del rules['ant_str']
95 del rules['con_str']
96 rules.reset_index(drop=True, inplace=True)
97
98 # Write association rules and metrics to file
99 rules.to_csv(outfile, sep="\t", index=False)
100
101
102 if __name__ == '__main__':
103 aparser = argparse.ArgumentParser()
104 aparser.add_argument("-i", "--inputs", dest="inputs", required=True)
105 aparser.add_argument("-y", "--infile", dest="infile", required=True)
106 aparser.add_argument("-o", "--outfile", dest="outfile", required=True)
107 aparser.add_argument("-s", "--support", dest="support", default=0.5)
108 aparser.add_argument("-c", "--confidence", dest="confidence", default=0.5)
109 aparser.add_argument("-l", "--lift", dest="lift", default=1.0)
110 aparser.add_argument("-v", "--conviction", dest="conviction", default=1.0)
111 aparser.add_argument("-t", "--length", dest="length", default=5)
112 args = aparser.parse_args()
113
114 main(args.inputs, args.infile, args.outfile,
115 min_support=float(args.support), min_confidence=float(args.confidence),
116 min_lift=float(args.lift), min_conviction=float(args.conviction), max_length=int(args.length))