@datadrivenconstruction/ddc-cwicr

Open Construction Cost Estimation Database — 55,719 work items, 27,672 resources, 85 fields, 9 languages with regional pricing.

Part of the Data Driven Construction ecosystem Live demo: openconstructionestimate.com

What is DDC CWICR?

Construction Work Items, Components & Resources (CWICR) — a comprehensive multilingual construction cost database designed for automated cost estimation from text, photos, PDFs, and CAD/BIM models.

Key Features

• 55,719 work items with full resource breakdowns
• 27,672 resources (materials, labor, equipment)
• 85 structured fields per record
• 9 languages with regional pricing
• Pre-computed OpenAI embeddings (3,072 dimensions) for semantic search
• Apache Parquet format for fast analytics

Supported Languages & Regions

| Code | Language | Region | Currency | | ---- | ----------- | ---------------- | -------- | | ar | Arabic | Dubai | AED | | de | German | Berlin | EUR | | en | English | Toronto | CAD | | es | Spanish | Barcelona | EUR | | fr | French | Paris | EUR | | hi | Hindi | Mumbai | INR | | pt | Portuguese | São Paulo | BRL | | ru | Russian | St. Petersburg | RUB | | zh | Chinese | Shanghai | CNY |

Installation

npm install @datadrivenconstruction/ddc-cwicr

This package includes all 9 parquet data files (~376 MB).

CLI Usage

# List all languages and data status
ddc-cwicr list

# Show details for a language
ddc-cwicr info en
ddc-cwicr info ru

# Get path to parquet file (for use in Python/scripts)
ddc-cwicr path en

# List all installed data paths
ddc-cwicr paths

Node.js API

const { DDCCWICR } = require('@datadrivenconstruction/ddc-cwicr');
const cwicr = new DDCCWICR();

// Get path to parquet file
const enPath = cwicr.getDataPath('en');
const ruPath = cwicr.getDataPath('ru');

// Check if data is available
cwicr.hasData('de'); // true

// List all languages
const languages = cwicr.listLanguages();

// Get only installed data paths
const paths = cwicr.getInstalledPaths();

// Language info
const info = cwicr.getLanguageInfo('fr');
// { code: 'fr', name: 'French', region: 'Paris', currency: 'EUR', file: '...' }

Python Usage

Use the parquet files directly with pandas:

import pandas as pd

# Get path from CLI
# $ ddc-cwicr path en
# C:\...\data\ENG_TORONTO_workitems_costs_resources_DDC_CWICR.parquet

df = pd.read_parquet('path/to/ENG_TORONTO_workitems_costs_resources_DDC_CWICR.parquet')

# Query work items
walls = df[df['work_item_name'].str.contains('wall', case=False)]
print(walls[['work_item_name', 'unit', 'total_cost_median']].head())

Data Schema (85 Fields)

Classification

• category, department, section, subsection

Work Item

• rate_code, work_item_name, unit, scope_flag

Resources

• material_code, material_name, material_unit, material_quantity, material_price

Labor

• workers, engineers, machinists, labor_hours, labor_wages

Machinery

• equipment_class, equipment_hours, equipment_rates, fuel_costs

Pricing

• total_cost_min, total_cost_max, total_cost_median

Embeddings

• embedding — OpenAI text-embedding-3-large (3,072 dimensions)

Cost Estimation Formula

Cost = Technology Norm × Regional Price

Example (Ceramic tile flooring, Berlin):
  Labor:     172 hrs/100m² × €17.95/hr    = €3,088.11
  Materials: 632 m²/100m²  × €5.02/m²     = €3,170.73
  Equipment: 1.67 hrs/100m² × €38.42/hr   = €64.18
  ─────────────────────────────────────────
  Total: €7,725.91 per 100m²

n8n Workflows

This package includes 4 production-ready n8n workflow templates:

1. Text-based cost estimation — Telegram bot with semantic search
2. Photo-based estimation — GPT-4 Vision analysis with auto-dimensioning
3. Universal bot — Text + Photo + PDF estimation via Telegram
4. CAD/BIM pipeline — 4D/5D cost estimation from Revit/IFC/DWG

Import from the workflows/ directory into your n8n instance.

Links

• GitHub Repository
• Live Demo
• DataDrivenConstruction.io

License

• Database: CC BY 4.0 — Attribution: "DDC CWICR by DataDrivenConstruction"
• Code: MIT