Real time predictive modelling tool for web app

I have a shopping site. I would like to build a simple predictive model to predict how likely an user will buy a certain product given a single visit (time spent on each page, where she came from, how many reviews she red etc.), when he is likely to leave a page etc.

What are the tools available for this task in Ruby on Rails, Python or Java? Do typical statistical modelling techniques such as logistic regressions, linear regressions work in this context?

I'd try to reduce the problem to a classification problem - and use Machine learning existing tools to get an answer.

Here are some steps you will need to take in order to do so:

1. Use your data and extract defined features out of it. A feature could be for example: number of pages visited or time spent on the web site, or anything else you can extract from your data.
2. Define what is the feature you want to "predict" (classify). A simple example could be: Buy a product (Let's start with buy any product, you can try to enhance it later on).
3. Create a training set. The training set contains as much classified examples as you can get. (for example: user i visited 5 different pages and spent 4 minutes, known classification: did not buy a product).
4. Given this information, you can run any existing classification algorithms, in trying to predict what a non classified user did, given her features alone.

A short list of some of the algorithms you can use for this:

1. SVM - not intuitive - but is considered by many the best classification algorithm available.
2. K Nearest neighbor - very intuitive and simple to program, and also the training set can be iteratively increased easily, but is usually a bad decision if the number of features is high.
3. Decision trees algorithms, especially C4.5 : allows very fast classification, and the resulting tree is intuitive and readable to humans.

I don't know about Ruby on rails or python tools, but in Java - there is an open source project called Weka, that has these classification algorithms, and more.

Note:

You can evaluate your algorithm and get your confusion matrix (Evaluation how much the algorithm was right and how much it was wrong, and how), by using cross-validation on the training set.