Omarrran's picture
Upload README.md
8655465 verified
metadata
license: apache-2.0
language:
  - en
pretty_name: 'CryptoXChain-500K: Multi-Network Blockchain Transaction Dataset'
size_categories:
  - 100K<n<1M
task_categories:
  - tabular-classification
  - tabular-regression
  - time-series-forecasting
  - feature-extraction
tags:
  - blockchain
  - cryptocurrency
  - bitcoin
  - bitcoin-cash
  - dash
  - dogecoin
  - ethereum-classic
  - transactions
  - utxo
  - cross-chain
  - financial
  - anomaly-detection
configs:
  - config_name: btc
    data_files:
      - split: train
        path: data/BTC_transactions_100k.jsonl
  - config_name: bch
    data_files:
      - split: train
        path: data/BCH_transactions_100k.jsonl
  - config_name: dash
    data_files:
      - split: train
        path: data/DASH_transactions_100k.jsonl
  - config_name: doge
    data_files:
      - split: train
        path: data/DOGE_transactions_100k.jsonl
  - config_name: etc
    data_files:
      - split: train
        path: data/ETC_transactions_100k.jsonl
  - config_name: utxo_all
    data_files:
      - split: train
        path: data/UTXO_ALL_400k.jsonl
  - config_name: all_chains
    data_files:
      - split: train
        path: data/ALL_CHAINS_500k.jsonl

πŸ”— CryptoXChain-500K: Multi-Network Blockchain Transaction Dataset

Dataset Summary

A large-scale, multi-chain blockchain transaction dataset containing 500,000 real transactions sampled across 5 cryptocurrency networks: Bitcoin (BTC), Bitcoin Cash (BCH), Dash (DASH), Dogecoin (DOGE), and Ethereum Classic (ETC). Structured for immediate use in machine learning, financial analytics, anomaly detection, graph neural networks, and cross-chain comparative research.

Collected across 10 independent sampling runs per chain, each capturing 10,000 transactions β€” totalling 100,000 transactions per chain.


πŸ“ Files in This Dataset

File Format Rows Description
BTC_transactions_100k.csv CSV 100,000 Bitcoin UTXO transactions
BCH_transactions_100k.csv CSV 100,000 Bitcoin Cash UTXO transactions
DASH_transactions_100k.csv CSV 100,000 Dash UTXO + PrivateSend transactions
DOGE_transactions_100k.csv CSV 100,000 Dogecoin UTXO transactions
ETC_transactions_100k.csv CSV 100,000 Ethereum Classic account-model transactions
UTXO_ALL_400k.csv CSV 400,000 Merged BTC+BCH+DASH+DOGE, unified schema
ALL_CHAINS_500k.csv CSV 500,000 Full dataset, all 5 chains combined
BTC_transactions_100k.jsonl JSONL 100,000 HuggingFace-native format
BCH_transactions_100k.jsonl JSONL 100,000 HuggingFace-native format
DASH_transactions_100k.jsonl JSONL 100,000 HuggingFace-native format
DOGE_transactions_100k.jsonl JSONL 100,000 HuggingFace-native format
ETC_transactions_100k.jsonl JSONL 100,000 HuggingFace-native format
UTXO_ALL_400k.jsonl JSONL 400,000 Merged UTXO JSONL
ALL_CHAINS_500k.jsonl JSONL 500,000 Full dataset JSONL

πŸ—οΈ Schema Reference

UTXO Chains (BTC, BCH, DASH, DOGE)

Column Type Description
hash string Unique transaction hash (64-char hex)
block_number int64 Block height at which tx was confirmed
block_timestamp string ISO 8601 UTC timestamp of block
chain string Chain identifier: BTC / BCH / DASH / DOGE
input_val float64 Total input value (native coin units)
output_val float64 Total output value (native coin units)
fee_val float64 Miner fee = input_val - output_val
fee_rate float64 fee_val / input_val (proportion)
input_count int64 Number of input UTXOs consumed
output_count int64 Number of output UTXOs created
is_coinbase bool True = block reward / miner transaction
is_privatesend bool True = DASH PrivateSend/CoinJoin tx (DASH only)
is_whale bool True = input_val > 99th percentile of chain
is_high_fee bool True = fee_rate > 1%

ETC (Ethereum Classic) β€” Account Model

Column Type Description
hash string Unique tx hash (0x-prefixed 64-char hex)
block_number int64 ETC block height
block_timestamp string ISO 8601 UTC timestamp
chain string Always "ETC"
from_address string Sender EOA or contract address
to_address string Recipient address; NULL = contract creation
value_etc float64 Transfer amount in ETC
gas int64 Gas limit set by sender
gas_price_gwei float64 Gas price in Gwei
receipt_status int64 1 = success, 0 = failed/reverted
tx_type string "transfer", "zero_value", "contract_creation"
is_failed bool True = receipt_status == 0
gas_efficiency float64 value_etc / gas (value per gas unit)

πŸ“ˆ Per-Chain Statistics

BTC β€” Bitcoin

  • Transactions: 100,000
  • Mean input: 2.279792 BTC | Median: 0.003457 BTC
  • Max input: 20,191.61 BTC (whale tx ~$1.8B)
  • Gini coefficient: 0.9857
  • Coinbase txns: 0.0320% | Whale txns: 1.00% | High fee: 9.91%
  • Avg inputs/tx: 2.17 | Avg outputs/tx: 2.73
  • Block range: Feb 27 2026 (~33 blocks)

BCH β€” Bitcoin Cash

  • Transactions: 100,000
  • Mean input: 16.509342 BCH | Median: 0.000139 BCH
  • Max input: 63,287.59 BCH
  • Gini coefficient: 0.9836
  • Coinbase txns: 0.4460% | High fee txns: 64.59% (dust micro-txns)
  • Avg inputs/tx: 2.01 | Avg outputs/tx: 2.30

DASH β€” Dash (Privacy Coin)

  • Transactions: 100,000
  • Mean input: 36.172438 DASH | Median: 0.800008 DASH
  • Max input: 15,390.83 DASH
  • Gini coefficient: 0.9372 (lowest β€” PrivateSend equalises amounts)
  • PrivateSend txns: 7.25% | Coinbase: 4.9190%
  • Avg inputs/tx: 3.79 (highest β€” CoinJoin multi-input)
  • Notable: Median exactly 0.8 DASH = CoinJoin denomination fingerprint

DOGE β€” Dogecoin

  • Transactions: 100,000
  • Mean input: 42,235.55 DOGE | Median: 0.013267 DOGE
  • Max input: 353,137,638 DOGE (largest tx in dataset!)
  • Gini coefficient: 0.9963 (highest β€” most unequal)
  • Coinbase txns: 0.2070% | High fee txns: 64.82%
  • Avg inputs/tx: 1.19 | Avg outputs/tx: 2.22

ETC β€” Ethereum Classic

  • Transactions: 100,000
  • Mean value: 102.97 ETC | Median: 0.383154 ETC
  • Max value: 500,000 ETC
  • Gini coefficient: ~0.97
  • Failed txns: 0.031% | Unique senders: 9,285
  • Mean gas: 53,267 units | Mean gas price: 5.85 Gwei
  • Block range: Nov 18–22 2020 (~5 day window)
  • Notable: Top sender in 22.4% of all sampled txns (exchange/contract)

⚠️ Known Issues & Caveats

  1. TEMPORAL MISMATCH: UTXO chains sampled Feb 27 2026; ETC sampled Nov 18–22 2020. Do NOT perform direct cross-chain temporal comparisons without acknowledging this gap.

  2. DASH NULLS: 7,249 rows have null input_val/output_val due to PrivateSend/CoinJoin privacy mixing. Flagged via is_privatesend=True. Exclude or handle separately in numeric analyses.

  3. COINBASE ZERO-INPUT: BTC (32), DOGE (207), BCH (446), DASH (4,919) rows have input_val=0 β€” these are block reward transactions, correctly flagged via is_coinbase=True.

  4. WHALE SKEW: All chains have Gini > 0.93. Mean values are unrepresentative of typical transactions. Always prefer median for central tendency.

  5. DOGE SCALE: DOGE values are in full DOGE (not satoshis). Max tx = 353M DOGE. Normalize before cross-chain numeric comparisons.

  6. ETC MODEL DIFFERENCE: ETC is account-model, not UTXO. Has no input_count/output_count. Use UTXO_ALL_400k files for UTXO-only analysis.

  7. SAMPLING BIAS: 10 runs Γ— 10,000 txns captures a narrow block window. Not a longitudinal time-series dataset.


πŸš€ Quick Start

pandas

import pandas as pd

# Single chain
btc = pd.read_csv("data/BTC_transactions_100k.csv")

# All UTXO chains merged
utxo = pd.read_csv("data/UTXO_ALL_400k.csv")

# Full 500K
full = pd.read_csv("data/ALL_CHAINS_500k.csv")

# Filter by chain
dash_only = utxo[utxo["chain"] == "DASH"]

# Clean UTXO analysis (exclude coinbase & PrivateSend)
clean = utxo[~utxo["is_coinbase"] & ~utxo["is_privatesend"]]

# Whale transactions only
whales = utxo[utxo["is_whale"] == True]

HuggingFace datasets library

from datasets import load_dataset

# Load single chain
btc = load_dataset("Omarrran/CryptoXChain_500K_Multi_Network_Blockchain_Transaction_Dataset", "btc")

# Load all UTXO chains merged
utxo = load_dataset("Omarrran/CryptoXChain_500K_Multi_Network_Blockchain_Transaction_Dataset", "utxo_all")

# Load ETC (account-model)
etc = load_dataset("Omarrran/CryptoXChain_500K_Multi_Network_Blockchain_Transaction_Dataset", "etc")

# Load full 500K all chains
full = load_dataset("Omarrran/CryptoXChain_500K_Multi_Network_Blockchain_Transaction_Dataset", "all_chains")

# Convert to pandas
df = btc["train"].to_pandas()

πŸ”¬ Suggested Research Tasks

Task Recommended File Key Columns
Fee prediction utxo_all input_val, input_count, output_count β†’ fee_val
Whale / anomaly detection utxo_all input_val, fee_rate, is_whale
Privacy tx classification dash is_privatesend, input_count, output_count
Cross-chain value comparison all_chains chain, input_val
Gas price prediction etc gas, gas_price_gwei, value_etc
Tx type classification etc tx_type
Block congestion modeling any block_number β†’ tx count per block
Gini / inequality research utxo_all input_val grouped by chain
Coinbase detection utxo_all is_coinbase
UTXO graph analysis utxo_all hash, input_count, output_count

πŸ“‹ Citation

@dataset{malik2026cryptoxchain,
  author    = {Malik, Omar Haq Nawaz},
  title     = {CryptoXChain-500K: Multi-Network Blockchain Transaction Dataset},
  year      = {2026},
  publisher = {HuggingFace},
  url       = {https://huggingface.co/datasets/Omarrran/CryptoXChain_500K_Multi_Network_Blockchain_Transaction_Dataset},
  chains    = {BTC, BCH, DASH, DOGE, ETC},
  total_txns= {500000},
  license   = {Apache-2.0}
}

πŸ‘€ Author

Omar Haq Nawaz Malik (HuggingFace: Omarrran) AI Engineer & NLP Researcher | BITS Pilani | Srinagar, Kashmir