knowledge of statistics/applied math necessary to do a ph d?

kruti · Joined: 29 Jun 2012 Posts: 3

I have completed my m tech and done my dissertation in DM (with Genetic Algorithms). The guidance was not that great, I did it superficially. I want to pursue a ph d in DM from UK. Is the knowledge of Applied Math, statistics or probability mandatory for it? I know the basic stuff, but when i sit to study the research papers, there are several equations and all which I dont understand.

Should i take up a course in math for this?

phil123 · Posted: Sun Jul 08, 2012 9:43 pm Post subject:

Hi,

There are some subfields of data mining that do not necessitate advanced maths or where the math is simple. For example, i'm thinking about some clustering algorithms or some association rule and frequent pattern mining algorithms that do not need much maths to understand.

Also, for projects that need more advanced math, not atl of them will ask you the same kind of math. For example, if you want to work on social network, maybe that some graph theory could help you depending on the topic. For some other projects using bayesian networks, maybe that statistics will help you better.

So I think that it depends mostly on your project. But of course, more math is always good to have.

kruti · Joined: 29 Jun 2012 Posts: 3

I want to do research in BigData. Would you please be kind enough to tell me what topics in mathematics do I need to brush up in order to understand this better?

Thanking you in anticipation.

phil123 · Posted: Mon Jul 09, 2012 2:15 pm Post subject:

BigData is very general. The term is usually used to refer to algorithms or data structure that are scalable for very large datasets.

But it does not tell what kind of algorithms or what kind of data.

So what kind of data you want to process? (image, multimedia, graphs, ... etc.) and what kind of algorithm or approach you want to use ? (bayesian networks, neural networks, etc...).

kruti · Joined: 29 Jun 2012 Posts: 3

Thanks for the prompt reply.

Here are the details:

Type of data - Image datasets
Approach - Genetic Algorithm

Thanking you in anticipation.

phil123 · Posted: Wed Jul 11, 2012 10:37 am Post subject:

Hi,

I will give you my opinion. For image processing, you probably need good mathematical skills, if you want to develop techniques for manipulating images. But if you use some already implemented library for manipulating the images, then your job may be easy. To see what kind of math is need for image processing, here is an example:

http://www.math-info.univ-paris5.fr/~gk/ANKARA/math_in_im_proc.pdf

For genetic algorithms, in my opinion, there are not much mathematics in this field. So if you do your research with an emphasis on genetic algorithms instead of image processing, there will probably be less math.

To make sure, you can check papers and books about genetic algorithms or the topic that you are interested in. This will help you to see how is the math level ;-)

Best,