nathan0 commited on
Commit
f6d6b95
1 Parent(s): 1298465

Upload folder using huggingface_hub

Browse files
Files changed (2) hide show
  1. README.md +181 -0
  2. sharechat.py +157 -0
README.md ADDED
@@ -0,0 +1,181 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ dataset_info:
3
+ config_name: first_domain
4
+ features:
5
+ - name: f_0
6
+ dtype: int64
7
+ - name: f_1
8
+ dtype: int8
9
+ - name: f_2
10
+ dtype: int32
11
+ - name: f_3
12
+ dtype: int32
13
+ - name: f_4
14
+ dtype: int32
15
+ - name: f_5
16
+ dtype: int32
17
+ - name: f_6
18
+ dtype: int32
19
+ - name: f_7
20
+ dtype: int32
21
+ - name: f_8
22
+ dtype: int32
23
+ - name: f_9
24
+ dtype: int32
25
+ - name: f_10
26
+ dtype: int32
27
+ - name: f_11
28
+ dtype: int32
29
+ - name: f_12
30
+ dtype: int32
31
+ - name: f_13
32
+ dtype: int32
33
+ - name: f_14
34
+ dtype: int32
35
+ - name: f_15
36
+ dtype: int32
37
+ - name: f_16
38
+ dtype: int32
39
+ - name: f_17
40
+ dtype: int32
41
+ - name: f_18
42
+ dtype: int32
43
+ - name: f_19
44
+ dtype: int32
45
+ - name: f_20
46
+ dtype: int32
47
+ - name: f_21
48
+ dtype: int32
49
+ - name: f_22
50
+ dtype: int32
51
+ - name: f_23
52
+ dtype: int32
53
+ - name: f_24
54
+ dtype: int32
55
+ - name: f_25
56
+ dtype: int32
57
+ - name: f_26
58
+ dtype: int32
59
+ - name: f_27
60
+ dtype: int32
61
+ - name: f_28
62
+ dtype: int32
63
+ - name: f_29
64
+ dtype: int32
65
+ - name: f_30
66
+ dtype: int32
67
+ - name: f_31
68
+ dtype: int32
69
+ - name: f_32
70
+ dtype: int32
71
+ - name: f_33
72
+ dtype: int8
73
+ - name: f_34
74
+ dtype: int8
75
+ - name: f_35
76
+ dtype: int8
77
+ - name: f_36
78
+ dtype: int8
79
+ - name: f_37
80
+ dtype: int8
81
+ - name: f_38
82
+ dtype: int8
83
+ - name: f_39
84
+ dtype: int8
85
+ - name: f_40
86
+ dtype: int8
87
+ - name: f_41
88
+ dtype: int8
89
+ - name: f_42
90
+ dtype: float32
91
+ - name: f_43
92
+ dtype: float32
93
+ - name: f_44
94
+ dtype: float32
95
+ - name: f_45
96
+ dtype: float32
97
+ - name: f_46
98
+ dtype: float32
99
+ - name: f_47
100
+ dtype: float32
101
+ - name: f_48
102
+ dtype: float32
103
+ - name: f_49
104
+ dtype: float32
105
+ - name: f_50
106
+ dtype: float32
107
+ - name: f_51
108
+ dtype: float32
109
+ - name: f_52
110
+ dtype: float32
111
+ - name: f_53
112
+ dtype: float32
113
+ - name: f_54
114
+ dtype: float32
115
+ - name: f_55
116
+ dtype: float32
117
+ - name: f_56
118
+ dtype: float32
119
+ - name: f_57
120
+ dtype: float32
121
+ - name: f_58
122
+ dtype: float32
123
+ - name: f_59
124
+ dtype: float32
125
+ - name: f_60
126
+ dtype: float32
127
+ - name: f_61
128
+ dtype: float32
129
+ - name: f_62
130
+ dtype: float32
131
+ - name: f_63
132
+ dtype: float32
133
+ - name: f_64
134
+ dtype: float32
135
+ - name: f_65
136
+ dtype: float32
137
+ - name: f_66
138
+ dtype: float32
139
+ - name: f_67
140
+ dtype: float32
141
+ - name: f_68
142
+ dtype: float32
143
+ - name: f_69
144
+ dtype: float32
145
+ - name: f_70
146
+ dtype: float32
147
+ - name: f_71
148
+ dtype: float32
149
+ - name: f_72
150
+ dtype: float32
151
+ - name: f_73
152
+ dtype: float32
153
+ - name: f_74
154
+ dtype: float32
155
+ - name: f_75
156
+ dtype: float32
157
+ - name: f_76
158
+ dtype: float32
159
+ - name: f_77
160
+ dtype: float32
161
+ - name: f_78
162
+ dtype: float32
163
+ - name: f_79
164
+ dtype: float32
165
+ - name: is_clicked
166
+ dtype: int8
167
+ - name: is_installed
168
+ dtype: int8
169
+ splits:
170
+ - name: train
171
+ num_bytes: 964563810
172
+ num_examples: 3240869
173
+ - name: validation
174
+ num_bytes: 43754365
175
+ num_examples: 147011
176
+ - name: test
177
+ num_bytes: 29159092
178
+ num_examples: 97972
179
+ download_size: 488593303
180
+ dataset_size: 1037477267
181
+ ---
sharechat.py ADDED
@@ -0,0 +1,157 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ """The dataset corresponds to roughly 10M random users who visited the ShareChat + Moj app over three months.
2
+ We have sampled each user's activity to generate 10 impressions corresponding to each user.
3
+ Our target variable is whether there was an install for an app by the user or not.
4
+ """
5
+
6
+
7
+ import csv
8
+ import json
9
+ import os
10
+ import glob
11
+ import polars as pl
12
+
13
+ import datasets
14
+
15
+
16
+ _CITATION = """\
17
+ @incollection{agrawal2023recsys,
18
+ title={RecSys Challenge 2023 Dataset: Ads Recommendations in Online Advertising},
19
+ author={Agrawal, Rahul and Brahme, Sarang and Maitra, Sourav and Srivastava, Abhishek and Irissappane, Athirai and Liu, Yong and Kalloori, Saikishore},
20
+ booktitle={Proceedings of the Recommender Systems Challenge 2023},
21
+ pages={1--3},
22
+ year={2023}
23
+ }
24
+ """
25
+
26
+ _DESCRIPTION = """\
27
+ The dataset corresponds to roughly 10M random users who visited the ShareChat + Moj app over three months.
28
+ We have sampled each user's activity to generate 10 impressions corresponding to each user.
29
+ Our target variable is whether there was an install for an app by the user or not.
30
+ """
31
+
32
+ _HOMEPAGE = "https://www.recsyschallenge.com/2023/"
33
+
34
+ # TODO: Add the licence for the dataset here if you can find it
35
+ _LICENSE = ""
36
+
37
+
38
+ _URLS = {
39
+ "first_domain": "https://cdn.sharechat.com/2a161f8e_1679936280892_sc.zip",
40
+ }
41
+
42
+
43
+ class Sharechat(datasets.ArrowBasedBuilder):
44
+ """The dataset for RecSys Challenge 2024."""
45
+
46
+ VERSION = datasets.Version("1.0.0")
47
+
48
+ # If you need to make complex sub-parts in the datasets with configurable options
49
+ # You can create your own builder configuration class to store attribute, inheriting from datasets.BuilderConfig
50
+ # BUILDER_CONFIG_CLASS = MyBuilderConfig
51
+
52
+ BUILDER_CONFIGS = [
53
+ datasets.BuilderConfig(name="first_domain", version=VERSION, description="This part of my dataset covers a first domain"),
54
+ ]
55
+
56
+ DEFAULT_CONFIG_NAME = "first_domain" # It's not mandatory to have a default configuration. Just use one if it make sense.
57
+
58
+ def _info(self):
59
+ if self.config.name == "first_domain": # This is the name of the configuration selected in BUILDER_CONFIGS above
60
+ # data format: f0,f1,...,f79, is_clicked, is_installed
61
+ # all_cat_columns = time_columns + cat_columns + binary_columns
62
+ id_columns = [('f_0', datasets.Value("int64"))]
63
+ time_columns = [('f_1', datasets.Value("int8"))]
64
+ cat_columns = [(f'f_{i}', datasets.Value("int32")) for i in range(2, 33)]
65
+ binary_columns = [(f'f_{i}', datasets.Value("int8")) for i in range(33, 42)]
66
+ dense_columns = [(f'f_{i}', datasets.Value("float")) for i in range(42, 80)]
67
+ other_columns = [('is_clicked', datasets.Value("int8"))]
68
+ label_columns = [('is_installed', datasets.Value("int8"))]
69
+ all_columns = id_columns + time_columns + cat_columns + binary_columns + dense_columns + other_columns + label_columns
70
+
71
+ features = datasets.Features(dict(all_columns))
72
+ else: # This is an example to show how to have different features for other domains
73
+ raise NotImplementedError("This configuration is not implemented yet")
74
+ return datasets.DatasetInfo(
75
+ # This is the description that will appear on the datasets page.
76
+ description=_DESCRIPTION,
77
+ # This defines the different columns of the dataset and their types
78
+ features=features, # Here we define them above because they are different between the two configurations
79
+ # If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
80
+ # specify them. They'll be used if as_supervised=True in builder.as_dataset.
81
+ # supervised_keys=("sentence", "label"),
82
+ # Homepage of the dataset for documentation
83
+ homepage=_HOMEPAGE,
84
+ # License for the dataset if available
85
+ license=_LICENSE,
86
+ # Citation for the dataset
87
+ citation=_CITATION,
88
+ )
89
+
90
+ def _split_generators(self, dl_manager):
91
+ # dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
92
+ # It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
93
+ # By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
94
+ urls = _URLS[self.config.name]
95
+ data_dir = dl_manager.download_and_extract(urls)
96
+
97
+ id_columns = [('f_0', pl.Int64)]
98
+ time_columns = [('f_1', pl.Int8)]
99
+ cat_columns = [(f'f_{i}', pl.Int32) for i in range(2, 33)]
100
+ binary_columns = [(f'f_{i}', pl.Int8) for i in range(33, 42)]
101
+ dense_columns = [(f'f_{i}', pl.Float32) for i in range(42, 80)]
102
+ other_columns = [('is_clicked', pl.Int8)]
103
+ label_columns = [('is_installed', pl.Int8)]
104
+ all_columns = id_columns + time_columns + cat_columns + binary_columns + dense_columns + other_columns + label_columns
105
+ self.dtypes_dict = dict(all_columns)
106
+
107
+ return [
108
+ datasets.SplitGenerator(
109
+ name=datasets.Split.TRAIN,
110
+ gen_kwargs={
111
+ "filepaths": sorted(glob.glob(os.path.join(data_dir, "sharechat_recsys2023_data", "train", "*.csv"))),
112
+ "date_start": 45,
113
+ "date_end": 64,
114
+ },
115
+ ),
116
+ datasets.SplitGenerator(
117
+ name=datasets.Split.VALIDATION,
118
+ gen_kwargs={
119
+ "filepaths": sorted(glob.glob(os.path.join(data_dir, "sharechat_recsys2023_data", "train", "*.csv"))),
120
+ "date_start": 65,
121
+ "date_end": 65,
122
+ },
123
+ ),
124
+ datasets.SplitGenerator(
125
+ name=datasets.Split.TEST,
126
+ gen_kwargs={
127
+ "filepaths": sorted(glob.glob(os.path.join(data_dir, "sharechat_recsys2023_data", "train", "*.csv"))),
128
+ "date_start": 66,
129
+ "date_end": 66,
130
+ },
131
+ ),
132
+ ]
133
+
134
+ def _generate_tables(self, filepaths, date_start, date_end):
135
+ # This method handles input defined in _split_generators to yield (key, table) tuples from the dataset.
136
+ # The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
137
+ for id, filepath in enumerate(filepaths):
138
+ if self.config.name == "first_domain":
139
+ pa_table = (
140
+ pl.scan_csv(filepath, separator='\t', dtypes=self.dtypes_dict)
141
+ .filter(pl.col("f_1").is_between(date_start, date_end))
142
+ .sort("f_1", descending=False)
143
+ .with_columns(pl.col("f_1").mod(7).alias("f_1"))
144
+ .collect().to_arrow()
145
+ )
146
+ yield id, pa_table
147
+ else:
148
+ raise NotImplementedError
149
+
150
+
151
+
152
+ # test the script
153
+ # datasets-cli test sharechat.py --save_info --all_configs
154
+ # share the dataset
155
+ # huggingface-cli login
156
+ # huggingface-cli repo create sharechat-dataset --type dataset
157
+ # huggingface-cli upload sharechat-dataset . . --repo-type dataset