| import csv |
| import os |
|
|
| import datasets |
|
|
| logger = datasets.logging.get_logger(__name__) |
|
|
| _DESCRIPTION = """ |
| The consumer compaint data set is derived from the consumer complaint database |
| for the purpose of benchmarking quantification / label shift algorithms. The |
| data set consists of records of compaints about consumer financial products and |
| services that the Consumer Financial Protection Bureau sent to companies for |
| response. Each record has a corresponding product / sub product field which can |
| be used as labels for text classification. |
| """ |
|
|
| _URLS = { |
| "ccd": "https://drive.google.com/u/0/uc?id=1UVCa-9oh5Oz5cPO7kQluY8am1pAUUbes&export=download", |
| } |
|
|
| _FIELDS = ["date", "product", "subproduct", "issue", "subissue", "text"] |
| _PRODUCT_CATEGORIES = [ |
| "Checking or savings account", |
| "Credit card or prepaid card", |
| "Credit reporting, credit repair services, or other personal consumer reports", |
| "Debt collection", |
| "Money transfer, virtual currency, or money service", |
| "Mortgage", |
| "Payday loan, title loan, or personal loan", |
| "Student loan", |
| "Vehicle loan or lease", |
| ] |
|
|
|
|
| class CcdConfig(datasets.BuilderConfig): |
| def __init__( |
| self, |
| training_files, |
| testing_files, |
| label_column="product", |
| label_classes=_PRODUCT_CATEGORIES, |
| **kwargs, |
| ): |
| super().__init__(version=datasets.Version("1.0.0", ""), **kwargs) |
| self.label_column = label_column |
| self.label_classes = label_classes |
| self.training_files = training_files |
| self.testing_files = testing_files |
|
|
|
|
| class Ccd(datasets.GeneratorBasedBuilder): |
| BUILDER_CONFIGS = [ |
| CcdConfig( |
| name="standard01", |
| description="", |
| training_files=[ |
| "201609.csv", |
| "201610.csv", |
| "201611.csv", |
| "201612.csv", |
| "201701.csv", |
| "201702.csv", |
| "201703.csv", |
| "201704.csv", |
| "201705.csv", |
| "201706.csv", |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| "201711.csv", |
| "201712.csv", |
| "201801.csv", |
| "201802.csv", |
| "201803.csv", |
| "201804.csv", |
| "201805.csv", |
| "201806.csv", |
| "201807.csv", |
| "201808.csv", |
| ], |
| testing_files=[ |
| "201809.csv", |
| "201810.csv", |
| "201811.csv", |
| "201812.csv", |
| "201901.csv", |
| "201902.csv", |
| ], |
| ), |
| CcdConfig( |
| name="standard02", |
| description="", |
| training_files=[ |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| "201711.csv", |
| "201712.csv", |
| "201801.csv", |
| "201802.csv", |
| "201803.csv", |
| "201804.csv", |
| "201805.csv", |
| "201806.csv", |
| "201807.csv", |
| "201808.csv", |
| "201809.csv", |
| "201810.csv", |
| "201811.csv", |
| "201812.csv", |
| "201901.csv", |
| "201902.csv", |
| "201903.csv", |
| "201904.csv", |
| "201905.csv", |
| "201906.csv", |
| ], |
| testing_files=[ |
| "201907.csv", |
| "201908.csv", |
| "201909.csv", |
| "201910.csv", |
| "201911.csv", |
| "201912.csv", |
| ], |
| ), |
| CcdConfig( |
| name="standard03", |
| description="", |
| training_files=[ |
| "201902.csv", |
| "201903.csv", |
| "201904.csv", |
| "201905.csv", |
| "201906.csv", |
| "201907.csv", |
| "201908.csv", |
| "201909.csv", |
| "201910.csv", |
| "201911.csv", |
| "201912.csv", |
| "202001.csv", |
| "202002.csv", |
| "202003.csv", |
| "202004.csv", |
| "202005.csv", |
| "202006.csv", |
| "202007.csv", |
| "202008.csv", |
| "202009.csv", |
| "202010.csv", |
| "202011.csv", |
| "202012.csv", |
| "202101.csv", |
| ], |
| testing_files=[ |
| "202102.csv", |
| "202103.csv", |
| "202104.csv", |
| "202105.csv", |
| "202106.csv", |
| "202107.csv", |
| ], |
| ), |
| CcdConfig( |
| name="majorshift01", |
| description="Training on 24 months before a major distribution shift and testing on 6 months after.", |
| training_files=[ |
| "201505.csv", |
| "201506.csv", |
| "201507.csv", |
| "201508.csv", |
| "201509.csv", |
| "201510.csv", |
| "201511.csv", |
| "201512.csv", |
| "201601.csv", |
| "201602.csv", |
| "201603.csv", |
| "201604.csv", |
| "201605.csv", |
| "201606.csv", |
| "201607.csv", |
| "201608.csv", |
| "201609.csv", |
| "201610.csv", |
| "201611.csv", |
| "201612.csv", |
| "201701.csv", |
| "201702.csv", |
| "201703.csv", |
| "201704.csv", |
| ], |
| testing_files=[ |
| "201705.csv", |
| "201706.csv", |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| ], |
| ), |
| CcdConfig( |
| name="majorshift02", |
| description="Training on 24 months before a major distribution shift and testing on 6 months after.", |
| training_files=[ |
| "201511.csv", |
| "201512.csv", |
| "201601.csv", |
| "201602.csv", |
| "201603.csv", |
| "201604.csv", |
| "201605.csv", |
| "201606.csv", |
| "201607.csv", |
| "201608.csv", |
| "201609.csv", |
| "201610.csv", |
| "201611.csv", |
| "201612.csv", |
| "201701.csv", |
| "201702.csv", |
| "201703.csv", |
| "201704.csv", |
| "201705.csv", |
| "201706.csv", |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| ], |
| testing_files=[ |
| "201711.csv", |
| "201712.csv", |
| "201801.csv", |
| "201802.csv", |
| "201803.csv", |
| "201804.csv", |
| ], |
| ), |
| CcdConfig( |
| name="majorshift03", |
| description="Training on 24 months before a major distribution shift and testing on 6 months after.", |
| training_files=[ |
| "201803.csv", |
| "201804.csv", |
| "201805.csv", |
| "201806.csv", |
| "201807.csv", |
| "201808.csv", |
| "201809.csv", |
| "201810.csv", |
| "201811.csv", |
| "201812.csv", |
| "201901.csv", |
| "201902.csv", |
| "201903.csv", |
| "201904.csv", |
| "201905.csv", |
| "201906.csv", |
| "201907.csv", |
| "201908.csv", |
| "201909.csv", |
| "201910.csv", |
| "201911.csv", |
| "201912.csv", |
| "202001.csv", |
| "202002.csv", |
| ], |
| testing_files=[ |
| "202003.csv", |
| "202004.csv", |
| "202005.csv", |
| "202006.csv", |
| "202007.csv", |
| "202008.csv", |
| ], |
| ), |
| CcdConfig( |
| name="skiptime01", |
| description="Training on 24 months before a major distribution shift, testing on 6 months after a period of time.", |
| training_files=[ |
| "201505.csv", |
| "201506.csv", |
| "201507.csv", |
| "201508.csv", |
| "201509.csv", |
| "201510.csv", |
| "201511.csv", |
| "201512.csv", |
| "201601.csv", |
| "201602.csv", |
| "201603.csv", |
| "201604.csv", |
| "201605.csv", |
| "201606.csv", |
| "201607.csv", |
| "201608.csv", |
| "201609.csv", |
| "201610.csv", |
| "201611.csv", |
| "201612.csv", |
| "201701.csv", |
| "201702.csv", |
| "201703.csv", |
| "201704.csv", |
| ], |
| testing_files=[ |
| "201805.csv", |
| "201806.csv", |
| "201807.csv", |
| "201808.csv", |
| "201809.csv", |
| "201810.csv", |
| ], |
| ), |
| CcdConfig( |
| name="skiptime02", |
| description="Training on 24 months before a major distribution shift, testing on 6 months after a period of time.", |
| training_files=[ |
| "201511.csv", |
| "201512.csv", |
| "201601.csv", |
| "201602.csv", |
| "201603.csv", |
| "201604.csv", |
| "201605.csv", |
| "201606.csv", |
| "201607.csv", |
| "201608.csv", |
| "201609.csv", |
| "201610.csv", |
| "201611.csv", |
| "201612.csv", |
| "201701.csv", |
| "201702.csv", |
| "201703.csv", |
| "201704.csv", |
| "201705.csv", |
| "201706.csv", |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| ], |
| testing_files=[ |
| "201811.csv", |
| "201812.csv", |
| "201901.csv", |
| "201902.csv", |
| "201903.csv", |
| "201904.csv", |
| ], |
| ), |
| CcdConfig( |
| name="skiptime03", |
| description="Training on 24 months before a major distribution shift, testing on 6 months after a period of time.", |
| training_files=[ |
| "201703.csv", |
| "201704.csv", |
| "201705.csv", |
| "201706.csv", |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| "201711.csv", |
| "201712.csv", |
| "201801.csv", |
| "201802.csv", |
| "201803.csv", |
| "201804.csv", |
| "201805.csv", |
| "201806.csv", |
| "201807.csv", |
| "201808.csv", |
| "201809.csv", |
| "201810.csv", |
| "201811.csv", |
| "201812.csv", |
| "201901.csv", |
| "201902.csv", |
| ], |
| testing_files=[ |
| "202003.csv", |
| "202004.csv", |
| "202005.csv", |
| "202006.csv", |
| "202007.csv", |
| "202008.csv", |
| ], |
| ), |
| |
| CcdConfig( |
| name="chrono01", |
| description="Training on 201507-201806 and testing on 201807-201906", |
| training_files=[ |
| "201507.csv", |
| "201508.csv", |
| "201509.csv", |
| "201510.csv", |
| "201511.csv", |
| "201512.csv", |
| "201601.csv", |
| "201602.csv", |
| "201603.csv", |
| "201604.csv", |
| "201605.csv", |
| "201606.csv", |
| "201607.csv", |
| "201608.csv", |
| "201609.csv", |
| "201610.csv", |
| "201611.csv", |
| "201612.csv", |
| "201701.csv", |
| "201702.csv", |
| "201703.csv", |
| "201704.csv", |
| "201705.csv", |
| "201706.csv", |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| "201711.csv", |
| "201712.csv", |
| "201801.csv", |
| "201802.csv", |
| "201803.csv", |
| "201804.csv", |
| "201805.csv", |
| "201806.csv", |
| ], |
| testing_files=[ |
| "201807.csv", |
| "201808.csv", |
| "201809.csv", |
| "201810.csv", |
| "201811.csv", |
| "201812.csv", |
| "201901.csv", |
| "201902.csv", |
| "201903.csv", |
| "201904.csv", |
| "201905.csv", |
| "201906.csv", |
| ], |
| ), |
| CcdConfig( |
| name="chrono02", |
| description="Training on 201607-201906 and testing on 201907-202006", |
| training_files=[ |
| "201607.csv", |
| "201608.csv", |
| "201609.csv", |
| "201610.csv", |
| "201611.csv", |
| "201612.csv", |
| "201701.csv", |
| "201702.csv", |
| "201703.csv", |
| "201704.csv", |
| "201705.csv", |
| "201706.csv", |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| "201711.csv", |
| "201712.csv", |
| "201801.csv", |
| "201802.csv", |
| "201803.csv", |
| "201804.csv", |
| "201805.csv", |
| "201806.csv", |
| "201807.csv", |
| "201808.csv", |
| "201809.csv", |
| "201810.csv", |
| "201811.csv", |
| "201812.csv", |
| "201901.csv", |
| "201902.csv", |
| "201903.csv", |
| "201904.csv", |
| "201905.csv", |
| "201906.csv", |
| ], |
| testing_files=[ |
| "201907.csv", |
| "201908.csv", |
| "201909.csv", |
| "201910.csv", |
| "201911.csv", |
| "201912.csv", |
| "202001.csv", |
| "202002.csv", |
| "202003.csv", |
| "202004.csv", |
| "202005.csv", |
| "202006.csv", |
| ], |
| ), |
| CcdConfig( |
| name="chrono03", |
| description="Training on 201707-202006 and testing on 202007-202106", |
| training_files=[ |
| "201707.csv", |
| "201708.csv", |
| "201709.csv", |
| "201710.csv", |
| "201711.csv", |
| "201712.csv", |
| "201801.csv", |
| "201802.csv", |
| "201803.csv", |
| "201804.csv", |
| "201805.csv", |
| "201806.csv", |
| "201807.csv", |
| "201808.csv", |
| "201809.csv", |
| "201810.csv", |
| "201811.csv", |
| "201812.csv", |
| "201901.csv", |
| "201902.csv", |
| "201903.csv", |
| "201904.csv", |
| "201905.csv", |
| "201906.csv", |
| "201907.csv", |
| "201908.csv", |
| "201909.csv", |
| "201910.csv", |
| "201911.csv", |
| "201912.csv", |
| "202001.csv", |
| "202002.csv", |
| "202003.csv", |
| "202004.csv", |
| "202005.csv", |
| "202006.csv", |
| ], |
| testing_files=[ |
| "202007.csv", |
| "202008.csv", |
| "202009.csv", |
| "202010.csv", |
| "202011.csv", |
| "202012.csv", |
| "202101.csv", |
| "202102.csv", |
| "202103.csv", |
| "202104.csv", |
| "202105.csv", |
| "202106.csv", |
| ], |
| ), |
| CcdConfig( |
| name="trainlow_testhigh01", |
| description="Training on low entropy label distribution splits and testing on high entropy splits", |
| training_files=[ |
| "201709.csv", |
| "202106.csv", |
| "202005.csv", |
| "202012.csv", |
| "202011.csv", |
| "202007.csv", |
| "202008.csv", |
| "202101.csv", |
| "202009.csv", |
| "202003.csv", |
| "202010.csv", |
| "202006.csv", |
| "202004.csv", |
| "202105.csv", |
| "202001.csv", |
| "202002.csv", |
| "201908.csv", |
| "202103.csv", |
| "202102.csv", |
| "201907.csv", |
| "202104.csv", |
| "201906.csv", |
| "201911.csv", |
| "201912.csv", |
| "201905.csv", |
| "201910.csv", |
| "201909.csv", |
| "201811.csv", |
| "201904.csv", |
| "201902.csv", |
| "201903.csv", |
| "201810.csv", |
| "201802.csv", |
| "201808.csv", |
| "201812.csv", |
| "201805.csv", |
| ], |
| testing_files=[ |
| "201710.csv", |
| "201803.csv", |
| "201807.csv", |
| "201809.csv", |
| "201804.csv", |
| "201707.csv", |
| "201708.csv", |
| "201901.csv", |
| "201711.csv", |
| "201806.csv", |
| "201705.csv", |
| "201712.csv", |
| ], |
| ), |
| CcdConfig( |
| name="trainlow_testhigh02", |
| description="Training on low entropy label distribution splits and testing on high entropy splits", |
| training_files=[ |
| "202004.csv", |
| "202105.csv", |
| "202001.csv", |
| "202002.csv", |
| "201908.csv", |
| "202103.csv", |
| "202102.csv", |
| "201907.csv", |
| "202104.csv", |
| "201906.csv", |
| "201911.csv", |
| "201912.csv", |
| "201905.csv", |
| "201910.csv", |
| "201909.csv", |
| "201811.csv", |
| "201904.csv", |
| "201902.csv", |
| "201903.csv", |
| "201810.csv", |
| "201802.csv", |
| "201808.csv", |
| "201812.csv", |
| "201805.csv", |
| "201710.csv", |
| "201803.csv", |
| "201807.csv", |
| "201809.csv", |
| "201804.csv", |
| "201707.csv", |
| "201708.csv", |
| "201901.csv", |
| "201711.csv", |
| "201806.csv", |
| "201705.csv", |
| "201712.csv", |
| ], |
| testing_files=[ |
| "201706.csv", |
| "201801.csv", |
| "201508.csv", |
| "201507.csv", |
| "201509.csv", |
| "201602.csv", |
| "201512.csv", |
| "201601.csv", |
| "201704.csv", |
| "201511.csv", |
| "201603.csv", |
| "201510.csv", |
| ], |
| ), |
| CcdConfig( |
| name="trainlow_testhigh03", |
| description="Training on low entropy label distribution splits and testing on high entropy splits", |
| training_files=[ |
| "201905.csv", |
| "201910.csv", |
| "201909.csv", |
| "201811.csv", |
| "201904.csv", |
| "201902.csv", |
| "201903.csv", |
| "201810.csv", |
| "201802.csv", |
| "201808.csv", |
| "201812.csv", |
| "201805.csv", |
| "201710.csv", |
| "201803.csv", |
| "201807.csv", |
| "201809.csv", |
| "201804.csv", |
| "201707.csv", |
| "201708.csv", |
| "201901.csv", |
| "201711.csv", |
| "201806.csv", |
| "201705.csv", |
| "201712.csv", |
| "201706.csv", |
| "201801.csv", |
| "201508.csv", |
| "201507.csv", |
| "201509.csv", |
| "201602.csv", |
| "201512.csv", |
| "201601.csv", |
| "201704.csv", |
| "201511.csv", |
| "201603.csv", |
| "201510.csv", |
| ], |
| testing_files=[ |
| "201701.csv", |
| "201605.csv", |
| "201607.csv", |
| "201612.csv", |
| "201604.csv", |
| "201606.csv", |
| "201608.csv", |
| "201703.csv", |
| "201610.csv", |
| "201611.csv", |
| "201609.csv", |
| "201702.csv", |
| ], |
| ), |
| CcdConfig( |
| name="trainhigh_testlow01", |
| description="Training on high entropy label distribution splits and testing on low entropy splits", |
| training_files=[ |
| "201710.csv", |
| "201803.csv", |
| "201807.csv", |
| "201809.csv", |
| "201804.csv", |
| "201707.csv", |
| "201708.csv", |
| "201901.csv", |
| "201711.csv", |
| "201806.csv", |
| "201705.csv", |
| "201712.csv", |
| "201706.csv", |
| "201801.csv", |
| "201508.csv", |
| "201507.csv", |
| "201509.csv", |
| "201602.csv", |
| "201512.csv", |
| "201601.csv", |
| "201704.csv", |
| "201511.csv", |
| "201603.csv", |
| "201510.csv", |
| "201701.csv", |
| "201605.csv", |
| "201607.csv", |
| "201612.csv", |
| "201604.csv", |
| "201606.csv", |
| "201608.csv", |
| "201703.csv", |
| "201610.csv", |
| "201611.csv", |
| "201609.csv", |
| "201702.csv", |
| ], |
| testing_files=[ |
| "201905.csv", |
| "201910.csv", |
| "201909.csv", |
| "201811.csv", |
| "201904.csv", |
| "201902.csv", |
| "201903.csv", |
| "201810.csv", |
| "201802.csv", |
| "201808.csv", |
| "201812.csv", |
| "201805.csv", |
| ], |
| ), |
| CcdConfig( |
| name="trainhigh_testlow02", |
| description="Training on high entropy label distribution splits and testing on low entropy splits", |
| training_files=[ |
| "201905.csv", |
| "201910.csv", |
| "201909.csv", |
| "201811.csv", |
| "201904.csv", |
| "201902.csv", |
| "201903.csv", |
| "201810.csv", |
| "201802.csv", |
| "201808.csv", |
| "201812.csv", |
| "201805.csv", |
| "201710.csv", |
| "201803.csv", |
| "201807.csv", |
| "201809.csv", |
| "201804.csv", |
| "201707.csv", |
| "201708.csv", |
| "201901.csv", |
| "201711.csv", |
| "201806.csv", |
| "201705.csv", |
| "201712.csv", |
| "201706.csv", |
| "201801.csv", |
| "201508.csv", |
| "201507.csv", |
| "201509.csv", |
| "201602.csv", |
| "201512.csv", |
| "201601.csv", |
| "201704.csv", |
| "201511.csv", |
| "201603.csv", |
| "201510.csv", |
| ], |
| testing_files=[ |
| "202004.csv", |
| "202105.csv", |
| "202001.csv", |
| "202002.csv", |
| "201908.csv", |
| "202103.csv", |
| "202102.csv", |
| "201907.csv", |
| "202104.csv", |
| "201906.csv", |
| "201911.csv", |
| "201912.csv", |
| ], |
| ), |
| CcdConfig( |
| name="trainhigh_testlow03", |
| description="Training on high entropy label distribution splits and testing on low entropy splits", |
| training_files=[ |
| "202004.csv", |
| "202105.csv", |
| "202001.csv", |
| "202002.csv", |
| "201908.csv", |
| "202103.csv", |
| "202102.csv", |
| "201907.csv", |
| "202104.csv", |
| "201906.csv", |
| "201911.csv", |
| "201912.csv", |
| "201905.csv", |
| "201910.csv", |
| "201909.csv", |
| "201811.csv", |
| "201904.csv", |
| "201902.csv", |
| "201903.csv", |
| "201810.csv", |
| "201802.csv", |
| "201808.csv", |
| "201812.csv", |
| "201805.csv", |
| "201710.csv", |
| "201803.csv", |
| "201807.csv", |
| "201809.csv", |
| "201804.csv", |
| "201707.csv", |
| "201708.csv", |
| "201901.csv", |
| "201711.csv", |
| "201806.csv", |
| "201705.csv", |
| "201712.csv", |
| ], |
| testing_files=[ |
| "201709.csv", |
| "202106.csv", |
| "202005.csv", |
| "202012.csv", |
| "202011.csv", |
| "202007.csv", |
| "202008.csv", |
| "202101.csv", |
| "202009.csv", |
| "202003.csv", |
| "202010.csv", |
| "202006.csv", |
| ], |
| ), |
| ] |
|
|
| def _info(self): |
| features = { |
| "date": datasets.Value("string"), |
| "id": datasets.Value("int32"), |
| "label": datasets.features.ClassLabel(names=self.config.label_classes), |
| "text": datasets.Value("string"), |
| } |
| return datasets.DatasetInfo( |
| description=_DESCRIPTION, |
| features=datasets.Features(features), |
| ) |
|
|
| def _split_generators(self, dl_manager): |
| downloaded_files = dl_manager.download_and_extract(_URLS) |
| logger.info(str(downloaded_files)) |
| dirname = downloaded_files["ccd"] |
| train_filepaths = tuple( |
| os.path.join(dirname, "ccd", fname) for fname in self.config.training_files |
| ) |
| test_filepaths = tuple( |
| os.path.join(dirname, "ccd", fname) for fname in self.config.testing_files |
| ) |
| return [ |
| datasets.SplitGenerator( |
| name=datasets.Split.TRAIN, |
| gen_kwargs={"filepaths": train_filepaths}, |
| ), |
| datasets.SplitGenerator( |
| name=datasets.Split.TEST, |
| gen_kwargs={"filepaths": test_filepaths}, |
| ), |
| ] |
|
|
| def _generate_examples(self, filepaths): |
| logger.info(f"generating examples from {len(filepaths)} files") |
| idx = 0 |
| for filepath in filepaths: |
| with open(filepath, encoding="utf-8") as f: |
| reader = csv.DictReader(f, fieldnames=_FIELDS) |
| for row in reader: |
| yield idx, { |
| "date": row["date"], |
| "id": idx, |
| "label": row[self.config.label_column], |
| "text": row["text"], |
| } |
| idx += 1 |
|
|
