Comment by maCDzP

I work in this field - below are my two cents.

I have a hard time seeing that this approach would work. The environment is very messy and there are many field tested methods that don't involve digging up the whole street. One example of such a method is pipe bursting.

There have been loads of research in using AI (neural nets etc.) to find and predict pipe bursts. But I believe simple linear regression is enough since my gut is telling me age and material is a good predictor.

Where I live there is a strong seasonality because the ground freezes over and during thawing we have a lot of leaks. This is a local issue since it depends on both weather and geological environment.