README.md

metadata

license: apache-2.0
task_categories:
  - text-classification
language:
  - en
tags:
  - data-preprocessing
  - automl
  - benchmarks
size_categories:
  - n<1K
  - 1K<n<10K
  - 10K<n<100K
  - 100K<n<1M
configs:
  - config_name: twenty_newsgroups
    data_files:
      - split: train
        path: twenty_newsgroups/train.csv
      - split: test
        path: twenty_newsgroups/test.csv
      - split: validation
        path: twenty_newsgroups/validation.csv
  - config_name: banking77
    data_files:
      - split: train
        path: banking77/train-*.parquet
      - split: test
        path: banking77/test-*.parquet
      - split: validation
        path: banking77/validation-*.parquet
  - config_name: trec
    data_files:
      - split: train
        path: trec/train-*.parquet
      - split: test
        path: trec/test-*.parquet
      - split: validation
        path: trec/validation-*.parquet
  - config_name: financial_phrasebank
    data_files:
      - split: train
        path: financial_phrasebank/train-*.parquet
      - split: test
        path: financial_phrasebank/test-*.parquet
      - split: validation
        path: financial_phrasebank/validation-*.parquet
  - config_name: MASSIVE
    data_files:
      - split: train
        path: MASSIVE/train-*.parquet
      - split: test
        path: MASSIVE/test-*.parquet
      - split: validation
        path: MASSIVE/validation-*.parquet

Data Preprocessing AutoML Benchmarks

This repository contains text classification datasets with known data quality issues for preprocessing research in AutoML.

Usage

Load a specific dataset configuration like this:

from datasets import load_dataset

# Example for loading the TREC dataset
dataset = load_dataset("MothMalone/data-preprocessing-automl-benchmarks", "trec")

Available Datasets

Below are the details for each dataset configuration available in this repository.

banking77

• Description: Dataset of online banking queries annotated with their corresponding intents.
• Data Quality Issue: N/A
• Classes: 77

trec

• Description: The Text REtrieval Conference (TREC) Question Classification dataset contains 5500 labeled questions in training set and another 500 for test set.
• Data Quality Issue: N/A
• Classes: 6

financial_phrasebank

• Description: Financial sentiment analysis dataset with phrases from financial news.
• Data Quality Issue: N/A
• Classes: 3

MASSIVE

• Description: Multilingual Amazon Slurp Synthetic Intent and Voice Evaluation dataset.
• Data Quality Issue: N/A
• Classes: 60

twenty_newsgroups

• Description: The 20 Newsgroups dataset is a collection of approximately 20,000 newsgroup documents, partitioned across 20 different newsgroups.
• Data Quality Issue: N/A
• Classes: 20