toolathon123/manufacturing-cost-optimization-2026Q3

Manufacturing Quarterly Cost Optimization Dataset (2026Q3)

Unified quarterly cost-analysis dataset for the manufacturing group, merged from the China / Japan / India factory datasets hosted on Hugging Face.

Contents

• 11,100 records (>= 10,000) covering 9 plants across 3 regions.
• Source datasets:
  • toolathon123/manufacturing-cn-energy-2026Q3 — China energy & raw material (4,200 rows)
  • toolathon123/manufacturing-jp-maintenance-2026Q3 — Japan maintenance & downtime (3,100 rows)
  • toolathon123/manufacturing-in-labor-2026Q3 — India labor & output (3,800 rows)

Columns

Column	Type	Description
timestamp	datetime	Record timestamp (UTC)
plant_id	string	Plant identifier, e.g. CN-SH, JP-TK, IN-MU
region	string	China, Japan or India
quarter	string	Reporting quarter, 2026Q3
shift	string	India: Morning / Evening / Night
machine_id	string	Japan: machine identifier
energy_kwh	float	China: energy consumption (kWh)
raw_material_kg	float	China: raw material consumption (kg)
maintenance_cost_usd	float	Japan: maintenance cost (USD)
labor_cost_usd	float	India: labor cost (USD)
downtime_min	float	Japan: downtime (minutes)
labor_hours	float	India: labor hours
output_units	float	Produced units
energy_cost	float	Derived: energy_kwh * 0.12
maintenance_cost	float	Derived: maintenance_cost_usd
labor_cost	float	Derived: labor_cost_usd
total_cost	float	Derived: energy + maintenance + labor
unit_cost	float	Derived: total_cost / output_units

Cleaning & aggregation rules

• timestamp parsed as datetime.
• Missing values in cost-related columns filled with 0.
• Per-row costs derived then aggregated per plant_id (see plant_cost_summary.csv).

Files

• data/train-00000-of-00001.parquet — unified cleaned dataset (11,100 rows)
• plant_cost_summary.csv — quarterly total cost & unit cost per plant
• cost_report_2026Q3.md — markdown cost report
• README.md — this card

Usage

from datasets import load_dataset

ds = load_dataset("toolathon123/manufacturing-cost-optimization-2026Q3", split="train")
df = ds.to_pandas()
print(df.head())