Analogical Modeling of Board Game Ratings
Description
In the winter of 2006, I took an Analogical Modeling class from Dr. Skousen. In Analogical Modeling (AM), you describe objects with many different variables and that object's outcome. These objects make up a dataset. AM looks through these variables and tries to find patterns for specific outcomes. Objects without outcomes can than be created as a testset and AM will predict the outcomes for these new objects. Classic examples of AM involve using AM to predict whether a mushroom will be poisonous or not or what team will win the Final four.
The beauty behind AM is that you don't have to know which variables are important for the outcome. AM sorts through all of these for you and finds patterns that you might not have even recognized.
I decided to use AM to help me make decisions on purchasing board games. I created a web site that took all the board games I owned, downloaded different variables about them from BoardGameGeek, and used RegEx to parse through the variables and output them into a dataset. I then used a similar method to build my testset of games I wanted to purchase. The dataset and testset were run through AM and then the results file read back in by another PHP script to output the predicated enjoyment of the games visually.