Font Size: 

In recent years several studies were conducted exploring state of the practice and different issues in web and multimedia systems development. One of them was performed in Croatia among over hundred web development companies. Data analysis in all studies were conducted using classical statistical methods.
This paper provides enriched research results of the Croatian study where data was analyzed with the advanced techniques like learning in Bayesian networks, clustering and a decision tree induction. Those techniques enabled us to identify new validity of the development processes and web project management in dependence on parameters such as development experience, web company size, type and complexity of the web system, estimated quality level etc. Finally, an intelligent model for web systems development is proposed which suggests desirable approaches to different web development phases, useful to both developers and clients.