This project is a coursework for Data mining and neural networks module at Univesity of Leicester, led by Prof. Alexander Gorban - http://www2.le.ac.uk/departments/mathematics/extranet/staff-material/staff-profiles/ag153
It takes on the question whether the outcome of elections is predictable given a set of 12 questions concerning some economical, political and inter-party dynamics. The aim is to, in the future, transform the exercise (assignment in attached file) into a proper neural network using a subset of the data as a training set.
Code is writen and run in R; for the report, package stargazer by Marek Hlavac was used for LaTeX tables and subsequently reformated.
The code is not generalised to accomodate choosing different number of questions - there are three functions for 1,2 and 3 best questions search. However, it can run on a different, possibly extended, dataset of the same kind.