CoolFace
Datasetpublic

abhinavdread/msme-payment-dispute-dataset

MSME Payment Dispute Dataset Description This dataset contains structured case-level information for MSME payment disputes in India. The dataset was constructed from structured extraction of: MSME arbitration awards Commercial court decisions Public legal case repositories All records are anonymized and structured for machine learning purposes. Dataset Size ~4,600 structured cases 3 outcome classes: win settlement escalation… See the full description on the dataset page: https://huggingface.co/datasets/abhinavdread/msme-payment-dispute-dataset.

sourceHugging Facemitupdated 7mo agoView on Hugging Face
0likes7downloads
Dataset Card

MSME Payment Dispute Dataset

Description

This dataset contains structured case-level information for MSME payment disputes in India.

The dataset was constructed from structured extraction of:

  • —MSME arbitration awards
  • —Commercial court decisions
  • —Public legal case repositories

All records are anonymized and structured for machine learning purposes.

Dataset Size

  • —~4,600 structured cases
  • —3 outcome classes:
  • —win
  • —settlement
  • —escalation

Features

ColumnDescription
claim_amountMonetary claim value
delay_daysPayment delay duration
buyer_typegovt / private
contract_presentWhether formal contract exists
industry_sectorSector classification
claim_imputedIndicator if claim was imputed
delay_imputedIndicator if delay was imputed
outcome_labelTarget outcome (win / settlement / escalation)

Preprocessing Steps

  • —Removed rare labels
  • —Removed invalid zero values
  • —Outcome-based delay imputation
  • —Model-based claim imputation
  • —Missingness flags added (claim_imputed, delay_imputed)

Intended Use

  • —Research in legal AI
  • —Tabular ML benchmarking
  • —MSME risk analysis
  • —Delay & claim prediction modeling
  • —Fairness & bias studies in legal outcomes

Limitations

  • —Structured data only (tabular)
  • —No raw legal document text included
  • —Synthetic augmentation applied in some cases
  • —Not an official government dataset
  • —Potential selection bias (only decided/arbitrated cases)