Size Estimation of Puzzle Games

Software game development industry is increasing day by day. Games development is different and more complicated than other software development models. This leads toward difficulty in the accurate estimation of size, cost and effort required for successful development of computer games. Many projects fail due to poor size estimation. In this research, we developed a model to estimate the size of puzzle games. This work is an extension of the board-games model for size estimation. Dataset used for this research is collected from open source code repositories. We analyzed the predictors already defined for the board-based size estimation model and identified a new predictor for size estimation of puzzle games. Applying SLR and MLR on the data set of puzzle games gave us promising results. The value of R² obtained is 0.93. Results of validation of the model are MMRE 0.29, PRED (30) is 0.80. Using the K- fold cross validation, the model was validated using K=6. Results show that 4 out of 6 models have MMRE less than 0.30.