keywordbarrage/jumia-scraper

Python Code:

import sqlite3
from datasets import load_dataset
import pandas as pd

# Method 1: Load via Hugging Face datasets library
dataset = load_dataset("keyword-barrage/enterprise-sku-depletion-telemetry")
print("Hub Sample Record:", dataset["train"][0])

# Method 2: Load zero-latency local pre-indexed SQLite deliverable
conn = sqlite3.connect("data/enterprise_feed.sqlite")
df = pd.read_sql_query("SELECT sku_id, price_usd, stock_level, depletion_velocity FROM sku_stream ORDER BY timestamp DESC LIMIT 10", conn)
print(df)

Dataset Card:

Dataset Summary

This repository contains an open 1,000-row verified benchmark sample of the Jumia SKU depletion feed. It is structured with pre-indexed SQLite and Snappy Parquet partitions for quantitative market intelligence research. This feed addresses the core enterprise problem of brittle scraping pipelines, where Cloudflare Bot Management and TLS/JA4 fingerprinting often result in 403 errors, destroying data continuity for quantitative analysts.

Our extraction protocol bypasses edge challenge loops via transport-level protocol parity, ensuring 99.9% verified pipeline availability for high-frequency inventory tracking.

Enterprise Deliverables Matrix

This dataset is provided in four professional formats optimized for quantitative data engineering:

• Pre-indexed SQLite (.sqlite): Instant local querying with pre-built indices on sku_id, price_delta, and timestamp.
• Snappy Parquet: Optimized for direct ingestion into Snowflake, Google BigQuery, and AWS Athena.
• Zero-Dependency Offline DataGrid (.html): Self-contained, interactive table for non-technical quantitative stakeholders to verify data quality.
• CDC JSONL Stream: Change Data Capture format designed for Enterprise RAG pipelines and vector database updates.

Dataset Schema

Column	Type	Nullable	Description
sku_id	String	No	Unique Jumia/Merchant product identifier
timestamp	DateTime	No	ISO-8601 capture time
price_usd	Float	No	Current retail price in USD
stock_level	Integer	Yes	Inventory count (if available)
velocity_depletion	Float	No	Calculated rate of SKU depletion
target_domain	String	No	Source regional domain (e.g., jumia.com.ng)
tls_ja4_signature	String	No	Fingerprint of the extraction session
delivery_status	String	No	Success/Failure status of the edge probe

Sample JSON Record:

{
  "sku_id": "JA506EL12345",
  "timestamp": "2026-09-24T14:30:00Z",
  "price_usd": 124.50,
  "stock_level": 42,
  "velocity_depletion": 0.85,
  "target_domain": "jumia.com.ng",
  "tls_ja4_signature": "1b2c3d4e5f6g",
  "delivery_status": "verified_success"
}

Data Quality & SLA

• Availability: 99.9% uptime for daily delta updates.
• Parity: Deterministic schema maintenance across all regional Jumia domains.
• Latency: Sub-120ms protocol execution benchmarks for edge-case resolution.
• Support: This benchmark provides an open window into our managed enterprise feed. For full-scale production requirements, we maintain a dedicated infrastructure SLA.