Re: Ph.D topics in data mining or big data
Date: October 17, 2018 09:45PM
In pattern mining, there is a lot of possible topics:
- new applications of algorithms
- designing faster algorithms for an existing pattern mining problem (you can choose a problem and find a way of improving the performance)
- designing a new problem for pattern mining (you need to think about some limitations of the current pattern mining problems and think about something new that you can add to these problems, and then design an algorithm and show that your new problem is useful. A simple way of creating a new problem is to combine two problems together. For example, if there is a problem of sequential rule mining and another problem of mining patterns in a stream, you could combine to do sequential rule mining in a stream. It would be a new problem. But you need to make sure the problem is not too easy and is useful.
Also, a good start is to check my SPMF data mining software. There is a lot of code for existing algorithms. If you start from an existing algorithm where you already have some code and datasets, you can save a lot of time. Also you can learn from the code.
Hope this is helpful
Best,
Philippe