1
$\begingroup$

As a ML beginner, I don't how to begin with this problem, or more generally, are there any typical steps to take when approaching a real problem?

If I have some domain relevant knowledge, then I could make use of it to analyze the data and do the modelling. But if I don't, then all I have is just the data, a matrix of input and a vector of response (if supervised).

My question is if I don't have domain knowledge, what to do before I randomly try out some learning algorithms and tune the parameters? Am I supposed to analyze the data per se to find out any clue and how?

$\endgroup$
2
$\begingroup$

Some Exploratory Data Analysis can help you to get some hints about the data you're dealing with. See: http://en.wikipedia.org/wiki/Exploratory_data_analysis Or, for some exploratory data analysis techniques, take a look: https://www.coursera.org/course/exdata

$\endgroup$
2
  • $\begingroup$ Thanks, I read some about EDA, so before modeling, doing some EDA which might give us some insights about the data, which might be useful to come up with the right model, right? $\endgroup$ – avocado Jun 4 '14 at 9:33
  • $\begingroup$ Exactly, it gives you an idea of how the data is structured and, by that, it can show you if a given model would be relevant. $\endgroup$ – Jundiaius Jun 4 '14 at 11:21

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.