The problem of predicting total sales, to print the right amount of books, is faced.
A methodology to analyse data related to book sales prediction is proposed.
A feature selection process is conducted to find out the main factors influencing sales.
Real world data is analysed using several data mining and visualisation techniques.
Obtained models are able to predict sales from pre-publication data.
Predictive models to be used as decision-aid tools for book publishers are presented.