0

I'm halfway (3 months out of 6) through my masters thesis and the problem involves using a few open source software libraries to the analyse data. But essentially the problem is that after the data has been analysed, which is basically fitting data into a function, and the performance is good, I was basically done. There was no original work and anyone could have done that. So my supervisor then wanted me to explain why it performed so well, but the problem is that the algorithm is basically a statistical model and well-documented, with no relation to the data, so I would be just regurgitating summaries and superficial explanation of how the algorithm works, and the source-code basically does what it says. I explain this to him, that this would be a dead-end because the algorithm works with any data, and it was like explaining why a camera took better photos of my car than my drawing, which is down to the camera itself not the car, and my supervisor wants to why the car performs so well with the camera.

I am just stuck and worried on how I was going to write a thesis on this. I have done nothing new and my supervisor still wants me to go deeper to find a link, even though there may not be one. Should I ask an academic supervisor for advice? Should I just stick it out and develop a weak link between those two? Should I swap projects to something more experimental where at least the results are original since it was self-measured?

|improve this question
  • What is the distinction between your supervisor and an "academic supervisor"? The main goal of a master thesis is usually getting the degree with the secondary goal of getting a good grade. All of this is normally decided by the advisor and can be achieved by following their guidance. – Roland Jan 9 at 8:40
  • A supervisor is my direct adviser who supervises me on the project. An academic supervisor is a more administrative level that handles general problems. – user102857 Jan 9 at 9:09
  • What is the goal of your thesis, precisely? What were you supposed to do/demonstrate/...? – C.F Jan 9 at 11:19
  • A few questions to ask yourself, perhaps: Why was the particular data being analyzed? Why was the particular method selected (over others, even)? What was the goal of the thesis, anyways (what was the question)? Did something surprise you about the data, the methods or the results? Why are the results important? – CKM Jan 9 at 19:19
2

I'm halfway...through my masters thesis and the problem involves using a few open source software libraries to the analyse data[, which I've completed already, what should I do for the next three months?]

You could analyse the data using different techniques, draw conclusions from your analysis, ...

[My supervisor wants me to] explain why it performed so well, but [this is well-known].

Perhaps your supervisor wants you to demonstrate that you understand the well-known method, i.e., to explain the statistical model, the algorithm, etc.

I am just stuck and worried on how I was going to write a thesis on this.

You haven't mentioned the requirements for your thesis (they vary between universities). Perhaps you can edit your question to expand upon this.

my supervisor still wants me to go deeper to find a link, even though there may not be one.

Discovering the absence of a link is a research contribution (albeit, such research isn't considered so highly)

Should I ask an [administrative] supervisor for advice?

I don't see how an administrator can help.

Should I just stick it out and develop a weak link between those two?

That's one direction. Alternatively, find a better direction (and negotiated with your supervisor that you'll study that).

Should I swap projects to something more experimental where at least the results are original since it was self-measured?

Your halfway, swapping projects would be risky. Perhaps you can perform experiments on your data?

|improve this answer

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.