Performance Statistics
- Creator: Open University of the Netherlands
- Publisher: Rage project
- Owner: Rage project email
This asset provides Java libraries for automatic server-side learning analytics interpretation.
This asset provides Java libraries for automatic server-side learning analytics and risk assessment. The libraries include classes for compiling a statistically validated group performance measurement as well as classes for comparing players to groups, comparing groups to other groups, and assessing whether a player has a low score compared to a group. All classes update score per datapoint and do not require intermediate access to a database, making them suitable for online big-data systems such as apache storm/trident.
This asset provides seven classes for representation and processing of performance analytics. Classes include a group performance resource class and classes for updating, assessing, and serializing/deserializing. These classes could be used to assess the performance of a student compared to a group on a specific game level. Alternatively they could be used to compare multiple groups and in that way compare also compare multiple levels (for instance for the purpose of balancing game level order during playtesting).
Required input is simply a Double representing a student's score on some subsection of a game (such as a level or task). The double should be of an interval value type (so no categories such as '1 = good / 2 = bad').
Output depends on the requested operations by class. The risk assessment produces a boolean result (student at risk = true/false) for student vs group comparison.
An integer output (0 = groups are equal, 1 = group one scores higher, 2 = group 2 scores higher) for group vs group comparison.
The serializer produces a JSON string for the group data.
Note! Please be aware that the accuracy of statistics is initially low. Accuracy increases with more datapoints and 250 datapoints (playthroughs) is recommended for excellent accuracy. Therefore the advised performance reference group for a game level is a group containing all students that have played a level so far (as opposed to using school class groups of small sizes such as 10-20 students as the main reference group).
Support
The component is available "as is" without warranties or conditions of any kind. The authors of the software are not planning to provide any kind of technical support to third parties on any issue they may find with the component.
Sep 1, 2017
English
https://github.com/rageappliedgame/PerformanceStatisticsAssetJavaLibraries
Learning analytics
statistics
player profiling
Other
Other
Java
1.0
This is the initial release of the performance statistics asset. Future releases will include more extensive analytics, use-case examples, and implementation examples
Under Development
https://github.com/rageappliedgame/PerformanceStatisticsAssetJavaLibraries
Apache 2.0 (Apache License 2.0)