What topic to choose for the thesis?

Question

data mining

Good afternoon.
I was faced with the question of choosing a topic for my thesis. I would very much like to do something useful and necessary, and not another thesis for the sake of a diploma. I study at the department of software, so the thesis should be closely related to mathematics and programming.
There are three direction vectors for the diploma:
1) Data Mining - only partly familiar. There is a lot of mathematics here and the topic is interesting to me.
2) Development for iOS - more or less familiar. If something is to be done, then it should contain a decent amount of mathematics.
3) Artificial intelligence for finding the path - this topic is my favorite. I did the visualization of pathfinding algorithms and would most gladly take on some interesting task related to this direction.
I really hope for your help. I would be very happy if at least tell me in which direction to drip. And if you have some specific task for me related to my preferences, then that will be just wonderful!

Answer 1

Life advice:
1. Choose a free topic and make a diploma "for fuck off"
2. Find something to your liking, an interesting project, and develop it.
Honestly, no one needs your diploma. What is expected of you is not practical implementation, but mountains of signed papers. To the defense, all this bureaucracy will exhaust you so much that even the topic will become disgusting.

Answer 2

Do you have any priorities for further development? Start from them.

Answer 3

There are yandex.maps, yandex-plugs...API-open.
You can play around in this field: building a route, forecasting traffic jams, optimizing traffic lights.
Another interesting topic with mathematics: colorization of b/w images and videos. With hints, without hints, with automatic search for hints on the Internet (search for images of a similar plot, Google seemed to offer some kind of such service).
Processing of images from satellites. Monitoring of forest, water resources (fires/floods), dynamics/forecast.

Answer 4

Write a running city app (runcity.org). So that the user enters points, and the application itself builds the shortest route. Even cooler, if you can take a picture of the legend, and the program itself will find all the addresses (OCR + NLP) and add them to the map (OSM or Yandex API).