Show HN: Harnessing the power of AI to identify probable faults in used cars

This tool does two things;

1. A neural network predicts probable faults in used vehicles based on the vehicle's make, model, mileage and age. The predictions are derived from previous MOT failures for vehicles of a similar condition (MOTs are an annual vehicle test in the UK to assure roadworthiness). You can enter in those details at https://avoidalemon.com

2. (Of limited use outside the UK) The tool provides a visual representation of mileage over time for anybody who has a UK registered vehicle. The odometer reading over time is plotted on a graph which makes it is easier to see inflections which are indicative of changes in usage/ownership. If the vehicle has had the mileage clocked then this would also be more apparent. You can enter any registration at https://avoidalemon.com/car-report .

This is not a replacement for mechanical expertise or even for consulting suitable motoring forums. What I am hoping this tool does is provide a steer on where issues may be present based on previous MOT failures.

For those of you interested in the underlying data, the UK Government has open sourced a lot of data pertaining to vehicles. You can read more about it here - https://dvsa.github.io/mot-history-api-documentation/ , and here - https://developer-portal.driver-vehicle-licensing.api.gov.uk... .

Fair warning should you want to try this yourself – the data, particularly the MOT API, is very messy and required a significant preparation. Mechanics enter data into the national database mostly by freetext and there is little in the way of standardisation. For example, the manufacturer “Peugeot” was spelled in at least 10 different ways and, rather hilariously, the MOT datasheet is occasionally used to insult the vehicle owner - https://www.mirror.co.uk/news/uk-news/two-incredibly-rude-mo...

Let me know what you think!