Small project aimed to find the best model for predicting values for a given set of data. I'm working on it in hope to better understand forecasting techniques and statistics, alongside with genetic algorithms and neural networks.
Chosen CSV file is loaded. The file is expected to contain numbers one-number-for-a-line. A population of several entities is created, based on random 'DNA' code generation, and on preseted codes. Each entity's purpose is to predict values of a given data set and learn during the process, finally to predict future values. Entity's 'DNA' describes used model: combination of regression forecast, ARIMA, and neural networks (with varying topologies and neuron types). Entity fitness is based on RMSE of produced forecast.