Master thesis topics [closed]

Question

I am looking for a thesis to complete my master, I am interested in Predictive Analytics in marketing, HR, management or financial subject, using Data Mining Application.

I have found a very interesting subject: "Predicting customer churn using decision tree" or either "Predicting employee turnover using decision tree", I looked around very hard but unfortunately couldn't find any relevant dataset to download (Telecommunication Customer churn Dataset ).

I would like to work on a similar subject using "Decision Tree Technique".

Please suggest some topics or project that would make for a good masters thesis subject.

Thanks.

Answer 1

This is the approach I took:

1. Find journals related to your field of studies
2. Skim through the proceedings, see if there are titles that catch your interest
3. Read the papers (carefully or globally) that seemed interesting
4. Carefully consider the approaches and whatever future suggestions they present in their papers
5. Think critically: What would you change? What do you want to find out? Don't limit yourself to data but rather orient from the perspective of research. Solutions for data might only become apparent when you know exactly what you want to examine.

I think this has advantages because these papers outline details regarding data as well -- perhaps you can use the same.

Present some papers and your idea to your prospective supervisor and he/she will make some suggestions. Researchers generally have a lot of knowledge about the possibilities and might even be curious about some things themselves.

Good luck! And enjoy.

Answer 2

First, talk to your thesis advisor before committing to a project. They know better than I do.

Secondly, just analyzing a new dataset using standard techniques doesn't make for a good masters thesis. Your project is expected to use some sort of novel approach.

With that said, I'd suggest that you start by reading up on existing decision tree techniques, learning why they work and what their flaws are, and try to find ways to overcome the flaws. Then, once you have your improvement, it should be relatively easy to find a dataset to apply it to.