Which data mining algorithm should I use to find optimum performance (in this case)

Question

I have a dataset that contians the following information, time of the day, day of the week, performance of the post. The post is a blog post made on a certain blog, performance is computed using the number of visits, commenets, etc. We are trying to find a correlation between the time of posting, day of posting and performance. I am inclined to use a clustering algorithm, but I am not sure how to go about this, what algorithm would you recommend and why ?

Answer 1

Giving an advice on general things like the choice of the method is usually not easy -- and even more so if there is no data and only the principles are concerned.

Nevertheless, put in usual terms, it seems as if you wanted a model f(time of day, day of the week) which outputs a prediction on the performance. For this, you basically can use any regression method in which you feed your measured data, such as neural networks, kernel regression, regression trees (CART), etc.

Moreover, in order to get a first graphical interpretation, you can also use a histogram where you choose some time-window (like a quarter of an hour) and attribute to it the average performance in that time-window.

As said, so far these are only general things -- I hope that helps nevertheless.