mstz commited on
Commit
e114ad7
1 Parent(s): 8e8780b

Upload bank.py

Browse files
Files changed (1) hide show
  1. bank.py +50 -68
bank.py CHANGED
@@ -1,6 +1,7 @@
1
  """Bank Dataset"""
2
 
3
  from typing import List
 
4
 
5
  import datasets
6
 
@@ -31,7 +32,7 @@ _BASE_FEATURE_NAMES = [
31
  "age",
32
  "job",
33
  "marital_status",
34
- "education",
35
  "has_defaulted",
36
  "account_balance",
37
  "has_housing_loan",
@@ -42,6 +43,31 @@ _BASE_FEATURE_NAMES = [
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"
@@ -124,77 +150,33 @@ class Bank(datasets.GeneratorBasedBuilder):
124
 
125
  yield row_id, data_row
126
 
127
- def preprocess(self, data: pandas.DataFrame, config: str = "income") -> pandas.DataFrame:
128
- data.drop("day", axis="columns", inplace=True)
129
- data.drop("contact", axis="columns", inplace=True)
130
- data.drop("duration", axis="columns", inplace=True)
131
- data.drop("poutcome", axis="columns", inplace=True)
132
-
133
- # discretize features
134
- data.loc[:, "education"] = data.education.apply(self.encode_education)
135
- data.loc[:, "loan"] = data.loan.apply(self.encode_yes_no)
136
- data.loc[:, "housing"] = data.housing.apply(self.encode_yes_no)
137
- data.loc[:, "default"] = data.default.apply(self.encode_yes_no)
138
-
139
- data.columns = _BASE_FEATURE_NAMES
140
-
141
- data.loc[:, "successfull_subscription"] = data.successfull_subscription.apply(lambda x: 0 if x == "no" else 1)
142
 
143
- for f in _BASE_FEATURE_NAMES:
144
- print(f, data[f].max(), data.dtypes[f])
 
 
 
145
 
146
- if config == "subscription":
147
  return data
148
  else:
149
  raise ValueError(f"Unknown config: {config}")
 
 
 
 
 
 
 
 
 
 
 
 
150
 
151
- def encoding_dictionaries(self):
152
- education_dic, yes_no_dic = self.education_encoding_dic(), self.yes_no_encoding_dic()
153
- education_data = [("education", education, code) for education, code in education_dic.items()]
154
- loan_data = [("loan", loan, code) for loan, code in yes_no_dic.items()]
155
- housing_data = [("housing", housing, code) for housing, code in yes_no_dic.items()]
156
- default_data = [("default", default, code) for default, code in yes_no_dic.items()]
157
- data = pandas.DataFrame(education_data + loan_data + housing_data + default_data,
158
- columns=["feature", "original_value", "encoded_value"])
159
-
160
  return data
161
-
162
- def encode_education(self, education):
163
- return self.education_encoding_dic()[education]
164
-
165
- def decode_education(self, code):
166
- return self.education_decoding_dic()[code]
167
-
168
- def education_decoding_dic(self):
169
- return {
170
- 0: "unknown",
171
- 1: "primary",
172
- 2: "secondary",
173
- 3: "tertiary"
174
- }
175
-
176
- def education_encoding_dic(self):
177
- return {
178
- "unknown": 0,
179
- "primary": 1,
180
- "secondary": 2,
181
- "tertiary": 3
182
- }
183
-
184
- def encode_yes_no(self, yes_no):
185
- return self.yes_no_encoding_dic()[yes_no]
186
-
187
- def decode_yes_no(self, code):
188
- return self.yes_no_decoding_dic()[code]
189
-
190
- def yes_no_decoding_dic(self):
191
- return {
192
- 0: "no",
193
- 1: "yes"
194
- }
195
-
196
- def yes_no_encoding_dic(self):
197
- return {
198
- "no": 0,
199
- "yes": 1
200
- }
 
1
  """Bank Dataset"""
2
 
3
  from typing import List
4
+ from functools import partial
5
 
6
  import datasets
7
 
 
32
  "age",
33
  "job",
34
  "marital_status",
35
+ "education_level",
36
  "has_defaulted",
37
  "account_balance",
38
  "has_housing_loan",
 
43
  "number_of_calls_before_this_campaign",
44
  "successfull_subscription"
45
  ]
46
+ _ENCODING_DICS = {
47
+ "education_level": {
48
+ "unknown": 0,
49
+ "primary": 1,
50
+ "secondary": 2,
51
+ "tertiary": 3
52
+ },
53
+ "has_personal_loan": {
54
+ "no": 0,
55
+ "yes": 1
56
+ },
57
+ "has_housing_loan": {
58
+ "no": 0,
59
+ "yes": 1
60
+ },
61
+ "has_defaulted": {
62
+ "no": 0,
63
+ "yes": 1
64
+ },
65
+ "successful_subscription": {
66
+ "no": 0,
67
+ "yes": 1
68
+ }
69
+ }
70
+
71
 
72
  DESCRIPTION = "Bank dataset for subscription prediction."
73
  _HOMEPAGE = "https://archive.ics.uci.edu/ml/datasets/bank+marketing"
 
150
 
151
  yield row_id, data_row
152
 
153
+ def preprocess(self, data: pandas.DataFrame, config: str = "subscription") -> pandas.DataFrame:
154
+ if config == "subscription":
155
+ data.drop("day", axis="columns", inplace=True)
156
+ data.drop("contact", axis="columns", inplace=True)
157
+ data.drop("duration", axis="columns", inplace=True)
158
+ data.drop("poutcome", axis="columns", inplace=True)
 
 
 
 
 
 
 
 
 
159
 
160
+ data.columns = _BASE_FEATURE_NAMES
161
+
162
+ for feature in _ENCODING_DICS:
163
+ encoding_function = partial(self.encode, feature)
164
+ data.loc[:, feature] = data[feature].apply(encoding_function)
165
 
 
166
  return data
167
  else:
168
  raise ValueError(f"Unknown config: {config}")
169
+
170
+ def encode(self, feature, value):
171
+ if feature in _ENCODING_DICS:
172
+ return _ENCODING_DICS[feature][value]
173
+ raise ValueError(f"Unknown feature: {feature}")
174
+
175
+ def encoding_dics(self):
176
+ data = [pandas.DataFrame([(feature, original, encoded) for original, encoded in d.items()])
177
+ for feature, d in _ENCODING_DICS.items()]
178
+ data = pandas.concat(data, axis="rows").reset_index()
179
+ data.drop("index", axis="columns", inplace=True)
180
+ data.columns = ["feature", "original_value", "encoded_value"]
181
 
 
 
 
 
 
 
 
 
 
182
  return data