Empowers construction companies with process automation, utilizing open source code blocks and solutions borrowed from diverse industries

One of the main tools for building pipelines Pandas is the most popular library for data manipulation


Generate charts from Excel

An example of process automation

   Manual process ~5 minutes

  Pipeline runtime: ~5 seconds

1. Extract: data opening

# Importing the necessary libraries 
# for data manipulation and plotting

import pandas as pd
import matplotlib.pyplot as plt

# Reading the Excel file into a DataFrame (df)
df = pd.read_excel('C:DDCConstruction-Budget.xlsx')

# Show DataFrame table 
2. Transform: grouping and visualization
# Grouping by 'Category', summing 
# the 'Amount', plotting the results, and adjusting the layout

ax = df.groupby('Category')['Amount'].sum().plot(kind='bar', figsize=(10, 5), color='skyblue', title='Expenses by Category', ylabel='Amount', rot=45, grid=True).get_figure()
3. Load: export
# Specifying the path for saving 
# the figure and saving the plot as a PNG file

file_path = "C:DDCexpenses_by_category.png"


PDF from Revit or IFC project

An example of process automation

   Manual process ~20 minutes

  Pipeline runtime: ~20 seconds

1. Extract: data opening

import converter as ddc
import pandas as pd
# Standalone conversion to flat formats without opening Revit or using API and Forge
project_csv = ddc.revit('C:rme_basic_sample.rvt')

# Importing Revit and IFC data
df = pd.read_csv('C:rme_basic_sample.csv')
2.1 Transform: grouping and visualization
# Grouping a Revit or IFC project by parameters
dftable = df.groupby('Type')['Volume'].sum()
2.2 Transform: displaying 
# Displaying a table as a graph
graph = dftable.plot(kind='barh', figsize=(20, 12))
# Save graph as PNG
graphtopng = graph.get_figure()
graphtopng.savefig('C:DDC_samplegraph_type.png', bbox_inches = 'tight')
3 Load: export
# Installing the library that allows generating PDF documents
!pip install fpdf
from fpdf import FPDF

# Creating a PDF document based on the parameters found
pdf = FPDF()
pdf.set_font('Arial', 'B', 16)
pdf.cell(190, 8, 'Grouping of the entire project by Type parameter', 2, 1, 'C')
pdf.image('C:DDC_samplegraph_type.png', w = 180,  link = '')

# Saving a document in PDF format
pdf.output('Report_DataDrivenConstruction.pdf', 'F')

ETL is an acronym for three key components in data processing: Extract, Transform, and Load:

  • Extract: This step involves collecting data from a variety of sources. This data can range in format and structure from images to databases.

  • Transform: In this step, the data is processed and transformed. This may include data cleansing, aggregation, validation, and any application of process logic. The purpose of transformation is to bring the data to the format required for the final system.

  • Load: The final step where the processed data is loaded into the target system, document or repository, such as a Data Warehouse, Data Lakehouse or database.

A traditional manual or semi-automated data process that replicates the ETL process involves a data manager or project manager who manually monitors the process and manually creates reports and documents on the process. Such traditional data processing methods take a significant amount of time in an environment where the workday is strictly limited to the time frame of 9:00 am to 5:00 pm.

In most cases, to automate such processes, companies buy ready-made ERP-like solutions, which are often customized to the company's individual desires and customized by an external developer, who ultimately determines the efficiency and effectiveness of the system and ultimately directly affects the business efficiency of the company that has purchased such a system.

In case a company is not ready to operate or buy a comprehensive ERP system where processes are performed in a semi-automated mode, one way or another the company's management will start automating the company's processes outside the ERP systems.

In the automated version of the same ETL workflow, the overall process looks like a modular code that starts with processing data and translating it into a open structured form. Once the structured data is received, various scripts or modules are automatically, on a schedule, run to check for changes, transform and send messages.

The reasons why I work more and
more with ETL-pipelines are simple

Simon Dilhas

CEO & Co-Founder
Abstract AG 


A few years ago Data-Driven Construction team showed me Jupyter Notebooks and I fell in love.

Since then I do simple data manipulations quicker with a few lines of code instead of manually clicking in Excel. As always it was a gradual process, first I automated repeating tasks. As my confidence and know-how grows, I do more and more.

Moreover, with the advent of Chat GPT, it became even easier, by describing the desired outcome I get the code snippets and just need to adapt them.

It's saves me a lot of time. E.g. I have a list of new sign-ups and I have my company CRM. Once a month I check if all the signups are in the CRM. Before this was a manual comparison of two different lists. Now it's a press of a button and I get the list with the new signups in the right format to import it back to the CRM. By investing 1h of programming I'm saving 12h of work per year.

I reduce failures dramatically. Because I did not like to compare two lists I did not do it as regularly as I should have done it. Moreover, I'm not good at comparing data, every phone call disrupted my work and I forgot to change a data entry. The consequences were that not everybody who should have gotten my newsletter got it. 

I feel like a god when I program. The more complex work becomes the less direct control over the results I have. But when programming I'm in absolute control of the result. I write a few lines of code and the computer executes them exactly as I tell him. When the result is not as I expected it to be, it's usually because of a logical shortcut on my side, the computer does exactly what I tell him. 



A DataFrame is a way of organizing data into a table very similar to the one you might see in Excel. In this table, the rows are individual records or entities, and the columns are the various characteristics or attributes of these item-entities.

If we have a table with information about a construction project, the rows can represent the individual entities-elements of the project and the attributes-columns can represent their categories, parameters, position or coordinates of the BoundingBox elements.


🚀 Efficient Data Management

DataFrames are optimized for handling large datasets, providing faster data manipulation.

🌐 Support for Heterogeneous Data

They can store different data types (like integers, strings, and floats) in various columns, ideal for real-world data.

🔍 Built-in Operations

DataFrames come equipped with numerous built-in methods for data filtering, sorting, and aggregating, simplifying complex data operations.

🔬 Ease of Data Exploration

Their tabular structure makes it easy to explore, analyze, and visualize data, aiding in quick data inspection and analysis.

🔗 Compatibility with Data Analysis Tools

They seamlessly integrate with various data analysis and visualization libraries, enhancing productivity in data science tasks.

📊 Advanced Data Integration

DataFrames easily interface with different data sources and formats, facilitating the integration and consolidation of diverse data sets.

Every month fresh solutions and news in our social channels

Don't miss the new solutions

📤 The data in Revit, IFC and DWG models is created by one specialist, but there are dozens of other people working with that data, so the first priority is to optimize the flow of data throughout the organization, which LLMs such as ChatGPT models can significantly help with.

⚡️ DataDrivenConstruction converter and LLM-like ChatGPT language models pave the way for efficient process automation in processing data from Revit and IFC projects, eliminating the need for tedious manual data entry and analysis.

Relying solely on manual labor in the era of automation presents significant challenges


Human interventions can lead to varied outcomes, often influenced by fatigue, oversight, or misunderstanding


Manual processes can’t easily scale to handle large volumes of work or complex tasks without proportionally increasing resources or time


Manual tasks, especially repetitive ones, can be significantly slower than automated processes, leading to inefficiencies and delays

Utilizing a pipeline in data processing provides substantial benefits


Streamline data operations for faster and optimized execution


Ensure uniform results across datasets with reproducible outcomes


Structured code flow for easy understanding, debugging, and maintenance