Datasets:
Upload 2 files
Browse files
README.md
ADDED
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
language:
|
3 |
+
- en
|
4 |
+
tags:
|
5 |
+
- compas
|
6 |
+
- tabular_classification
|
7 |
+
- binary_classification
|
8 |
+
pretty_name: Bank
|
9 |
+
size_categories:
|
10 |
+
- 1K<n<10K
|
11 |
+
task_categories: # Full list at https://github.com/huggingface/hub-docs/blob/main/js/src/lib/interfaces/Types.ts
|
12 |
+
- tabular-classification
|
13 |
+
configs:
|
14 |
+
- encoding
|
15 |
+
- subscription
|
16 |
+
---
|
17 |
+
# Bank
|
18 |
+
The [Bank dataset](https://archive.ics.uci.edu/ml/datasets/bank+marketing) is cool.
|
bank.py
ADDED
@@ -0,0 +1,194 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
"""Bank Dataset"""
|
2 |
+
|
3 |
+
from typing import List
|
4 |
+
|
5 |
+
import datasets
|
6 |
+
|
7 |
+
import pandas
|
8 |
+
|
9 |
+
|
10 |
+
VERSION = datasets.Version("1.0.0")
|
11 |
+
_ORIGINAL_FEATURE_NAMES = [
|
12 |
+
"age",
|
13 |
+
"job",
|
14 |
+
"marital",
|
15 |
+
"education",
|
16 |
+
"default",
|
17 |
+
"balance",
|
18 |
+
"housing",
|
19 |
+
"loan",
|
20 |
+
"contact",
|
21 |
+
"day",
|
22 |
+
"month",
|
23 |
+
"duration",
|
24 |
+
"campaign",
|
25 |
+
"pdays",
|
26 |
+
"previous",
|
27 |
+
"poutcome",
|
28 |
+
"y"
|
29 |
+
]
|
30 |
+
_BASE_FEATURE_NAMES = [
|
31 |
+
"age",
|
32 |
+
"job",
|
33 |
+
"marital_status",
|
34 |
+
"education",
|
35 |
+
"has_defaulted",
|
36 |
+
"account_balance",
|
37 |
+
"has_housing_loan",
|
38 |
+
"has_personal_loan",
|
39 |
+
"month_of_last_contact",
|
40 |
+
"number_of_calls_in_ad_campaign",
|
41 |
+
"days_since_last_contact_of_previous_campaign",
|
42 |
+
"number_of_calls_before_this_campaign",
|
43 |
+
"successfull_subscription"
|
44 |
+
]
|
45 |
+
|
46 |
+
DESCRIPTION = "Bank dataset for subscription prediction."
|
47 |
+
_HOMEPAGE = "https://archive.ics.uci.edu/ml/datasets/bank+marketing"
|
48 |
+
_URLS = ("https://huggingface.co/datasets/mstz/bank/raw/main/bank-full.csv")
|
49 |
+
_CITATION = """"""
|
50 |
+
|
51 |
+
# Dataset info
|
52 |
+
urls_per_split = {
|
53 |
+
"train": "https://huggingface.co/datasets/mstz/bank/raw/main/bank-full.csv",
|
54 |
+
}
|
55 |
+
features_types_per_config = {
|
56 |
+
"encoding": {
|
57 |
+
"feature": datasets.Value("string"),
|
58 |
+
"original_value": datasets.Value("string"),
|
59 |
+
"encoded_value": datasets.Value("int8"),
|
60 |
+
},
|
61 |
+
|
62 |
+
"subscription": {
|
63 |
+
"age": datasets.Value("int64"),
|
64 |
+
"job": datasets.Value("string"),
|
65 |
+
"marital_status": datasets.Value("string"),
|
66 |
+
"education": datasets.Value("int8"),
|
67 |
+
"has_defaulted": datasets.Value("int8"),
|
68 |
+
"account_balance": datasets.Value("int16"),
|
69 |
+
"has_housing_loan": datasets.Value("int8"),
|
70 |
+
"has_personal_loan": datasets.Value("int8"),
|
71 |
+
"month_of_last_contact": datasets.Value("string"),
|
72 |
+
"number_of_calls_in_ad_campaign": datasets.Value("string"),
|
73 |
+
"days_since_last_contact_of_previous_campaign": datasets.Value("int16"),
|
74 |
+
"number_of_calls_before_this_campaign": datasets.Value("int16"),
|
75 |
+
"successfull_subscription": datasets.ClassLabel(num_classes=2, names=("no", "yes")),
|
76 |
+
}
|
77 |
+
|
78 |
+
}
|
79 |
+
features_per_config = {k: datasets.Features(features_types_per_config[k]) for k in features_types_per_config}
|
80 |
+
|
81 |
+
|
82 |
+
class BankConfig(datasets.BuilderConfig):
|
83 |
+
def __init__(self, **kwargs):
|
84 |
+
super(BankConfig, self).__init__(version=VERSION, **kwargs)
|
85 |
+
self.features = features_per_config[kwargs["name"]]
|
86 |
+
|
87 |
+
|
88 |
+
class Bank(datasets.GeneratorBasedBuilder):
|
89 |
+
# dataset versions
|
90 |
+
DEFAULT_CONFIG = "subscription"
|
91 |
+
BUILDER_CONFIGS = [
|
92 |
+
BankConfig(name="encoding",
|
93 |
+
description="Encoding dictionaries for discrete features."),
|
94 |
+
BankConfig(name="subscription",
|
95 |
+
description="Bank binary classification for client subscription."),
|
96 |
+
]
|
97 |
+
|
98 |
+
|
99 |
+
def _info(self):
|
100 |
+
if self.config.name not in features_per_config:
|
101 |
+
raise ValueError(f"Unknown configuration: {self.config.name}")
|
102 |
+
|
103 |
+
info = datasets.DatasetInfo(description=DESCRIPTION, citation=_CITATION, homepage=_HOMEPAGE,
|
104 |
+
features=features_per_config[self.config.name])
|
105 |
+
|
106 |
+
return info
|
107 |
+
|
108 |
+
def _split_generators(self, dl_manager: datasets.DownloadManager) -> List[datasets.SplitGenerator]:
|
109 |
+
downloads = dl_manager.download_and_extract(urls_per_split)
|
110 |
+
|
111 |
+
return [
|
112 |
+
datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"filepath": downloads["train"]}),
|
113 |
+
]
|
114 |
+
|
115 |
+
def _generate_examples(self, filepath: str):
|
116 |
+
data = pandas.read_csv(filepath, sep=";")
|
117 |
+
data = self.preprocess(data, config=self.config.name)
|
118 |
+
|
119 |
+
for row_id, row in data.iterrows():
|
120 |
+
data_row = dict(row)
|
121 |
+
|
122 |
+
yield row_id, data_row
|
123 |
+
|
124 |
+
def preprocess(self, data: pandas.DataFrame, config: str = "income") -> pandas.DataFrame:
|
125 |
+
data.drop("day", axis="columns", inplace=True)
|
126 |
+
data.drop("contact", axis="columns", inplace=True)
|
127 |
+
data.drop("duration", axis="columns", inplace=True)
|
128 |
+
data.drop("poutcome", axis="columns", inplace=True)
|
129 |
+
|
130 |
+
data.columns = _BASE_FEATURE_NAMES
|
131 |
+
|
132 |
+
# discretize features
|
133 |
+
data.loc[:, "education"] = data.education.apply(self.encode_education)
|
134 |
+
data.loc[:, "loan"] = data.loan.apply(self.encode_yes_no)
|
135 |
+
data.loc[:, "housing"] = data.housing.apply(self.encode_yes_no)
|
136 |
+
data.loc[:, "default"] = data.default.apply(self.encode_yes_no)
|
137 |
+
|
138 |
+
if config == "encoding":
|
139 |
+
return self.encoding_dictionaries()
|
140 |
+
elif config == "subscription":
|
141 |
+
return data
|
142 |
+
else:
|
143 |
+
raise ValueError(f"Unknown config: {config}")
|
144 |
+
|
145 |
+
def encoding_dictionaries(self):
|
146 |
+
education_dic, binary_dic = self.education_encoding_dic(), self.binary_encoding_dic()
|
147 |
+
education_data = [("education", education, code) for education, code in education_dic.items()]
|
148 |
+
loan_data = [("loan", loan, code) for loan, code in binary_dic.items()]
|
149 |
+
housing_data = [("housing", housing, code) for housing, code in binary_dic.items()]
|
150 |
+
default_data = [("default", default, code) for default, code in binary_dic.items()]
|
151 |
+
data = pandas.DataFrame(education_data, loan_data + housing_data + default_data,
|
152 |
+
columns=["feature", "original_value", "encoded_value"])
|
153 |
+
|
154 |
+
return data
|
155 |
+
|
156 |
+
def encode_education(self, education):
|
157 |
+
return self.education_encoding_dic()[education]
|
158 |
+
|
159 |
+
def decode_education(self, code):
|
160 |
+
return self.education_decoding_dic()[code]
|
161 |
+
|
162 |
+
def education_decoding_dic(self):
|
163 |
+
return {
|
164 |
+
0: "unknown",
|
165 |
+
1: "primary",
|
166 |
+
2: "secondary",
|
167 |
+
3: "tertiary"
|
168 |
+
}
|
169 |
+
|
170 |
+
def education_encoding_dic(self):
|
171 |
+
return {
|
172 |
+
"unknown": 0,
|
173 |
+
"primary": 1,
|
174 |
+
"secondary": 2,
|
175 |
+
"tertiary": 3
|
176 |
+
}
|
177 |
+
|
178 |
+
def encode_yes_no(self, yes_no):
|
179 |
+
return self.yes_no_encoding_dic()[yes_no]
|
180 |
+
|
181 |
+
def decode_yes_no(self, code):
|
182 |
+
return self.yes_no_decoding_dic()[code]
|
183 |
+
|
184 |
+
def yes_no_decoding_dic(self):
|
185 |
+
return {
|
186 |
+
0: "no",
|
187 |
+
1: "yes"
|
188 |
+
}
|
189 |
+
|
190 |
+
def yes_no_encoding_dic(self):
|
191 |
+
return {
|
192 |
+
"no": 0,
|
193 |
+
"yes": 1
|
194 |
+
}
|