Predictive Maintenance Software

Hi , I am planning to make an AI predictive maintenance software for private jets. In order to pitch , i need to have some data to use it for my prototype . While I have none . So how do i get an access to it in real time other than getting them from operators

Kaggle and GitHub are the source where you could find some of the data but GitHub would have privacy issue concern, so when you try searching on these portals make sure check it the data available is public or private.

other best source would be simply Google with the data you are looking for which will provide you further links or details on where you can find data related to your interest.


Thank you Deepti , I will look into this . I appreciate you for the advise .

+1 to that

And @Adnan_Ahmed, look at who else is already doing this…aircraft predictive maintenance is not a new idea. See for example

Thank you Sir , I have been taking other companies into account as well. The challenge remains in getting real time data . I have been working as a consultant in private aviation , I am thinking to reach out to plane owners and lets see how I can get things moving forward from there .

You don’t actually need real time data to develop and validate a model, do you? Seems to me that you could decouple or defer that requirement until you have something compelling to show people. Think ahead about how to monetize the value of your service; who would pay, why, and how (much)

You’re actually right… I’d proceed that way. I have prepared a case study and figured out how much companies loose because of this mechanical/technical breakdown. Even if we are accurate for 30 percent of the time we will still save a huge sum of money and on the other hand with practise and monitoring the models we can definitely make the model better . Then ramp up from there. Regarding costs , I’ll need to ascertain things and then price the software at a nominal rate so that it’s a win-win situation .

1 Like

Regarding model validation , i thought it could have been a good idea to have Atleast something solid for the prototype . It’s okay , I’ll still try to convince companies out there…

When I worked in healthcare (hospital readmissions) we trained the models with historical data. How else could we validate who in fact was readmitted at a later date? Once the model was trained, we connected it to current (still not real time) data to inform active care decisions. I never worked on aircraft data, but it seems the same approach pertains. Maybe I’m missing something.

As @Deepti_Prasad suggested above…

Has a dataset. Might explore where it came from.