Retrieve Dataset for Machine Learning
Opened this issue · 3 comments
At the beginning of each adaptive level, the k means algorithm is run using a hard-coded set of data in . This is not an ideal implementation.
The desired operation of the adaptation should be to pull the data from a database when the game starts up, after the student has indicated their identification number and course code.
The key features that should be used for this are failures per level and time taken to successfully complete the level, starting from the first attempt (both normalized). The 3 k means groups are sorted based on their failure rates.
Hard-coded data set is found in GameController.cs.
Instead of number of failed attempts at a level, we should consider using # of failed bugcatchers as the categorization determiner / combine that with the time taken
RoboBUG now pulls the data from the database. Using the TimeEnded/TimeStarted to find elapsed time and also amount of failed tool uses to determine the ML categorization.