QGIS we can easily analyze and model land use/land cover change. Using the MOLUSCE plugin we can evaluate the total change area at the land cover class level, model change probability, simulate change using cellular automata and perform model validation. One of the most interesting features of MOLUSCE is that it allows modeling transition potential using four different techniques: neural networks, logistic regression, multi-criteria evaluation, or weights of evidence.
Use MOLUSCE (Modules for Land Use Change Simulations)Plugin it provides a set of algorithms for land use change simulations such as ANN, LR, WoE, MCE as well as validation using kappa statistics.
Modules for Land Use Change Simulations is the best tool because accurate prediction requires you to include drivers of change. The above tool does it. Markov Model is another option for your work. search your web
When you want to predict evolution it would be advisable to use entropy as measures of complexity and as means of prediction use AI. The methodology is described in our paper on prediction of Torsades de Pointes arrhythmia. It is a very detailed description.