A machine learning (ML) approach can significantly increase the cost accuracy of estimates for organisations in the water industry with lots of historical project data. We’ve created a machine learning cost estimating tool that uses algorithms to detect relationships between the many possible different asset characteristics or data which our cost experts gather from historical projects.
Our tool creates patterns from the data by analysing correlations within entire asset datasets. This allows us to factor all available project scope into each asset, removing any subjectivity and assumptions. Ultimately this helps us make more accurate cost predictions faster, and with less resources, than traditional methods.
We’ve trialled our approach using actual costs from several hundred water pipe-laying projects, leading to final output costs 50 percent more accurate than those generated using traditional estimating approaches. Increased accuracy has the advantages of more accurate investment decisions and increased confidence in target cost setting for both client and contractor.