Data science meets standardized game learning analytics
1. Data science meets standardized
game learning analytics
Cristina Alonso-Fernández, Antonio Calvo-Morata, Manuel Freire,
Iván Martínez-Ortiz, Baltasar Fernández-Manjón
EDUCON 2021
2. Serious Games
Serious Games: main purpose other than entertainment
(learning, raising awareness, changing players’ attitudes)
Multiple benefits but still low adoption in education:
➔ Need for clear evidences for game validation and to
evaluate players
➔ Lack of infrastructure to analyze interaction data
➔ Complexity for educators and non data experts
➔ Cost of new ad-hoc personalized analysis for each game
3. Game Learning Analytics
Game Learning Analytics: collection and
analysis of data from serious games
➔ evaluate and improve games
➔ provide information that helps
educators and other stakeholders (e.g.
designers)
➔ evidence-based decisions for
institutions
➔ games that provide a more authentic
learning experience for students
play
collection & analysis
reports
information &
feedback
4. Standardizing interaction data collection: xAPI-SG
The Experience API for Serious Games
Profile (xAPI-SG) defines the common set
of interactions in serious games.
● Completables: players can start,
progress and complete
● Accessibles: to access or skip
● Alternatives: decisions in the game, to
choose among several options
● GameObjects: to interact with or use
Standardize in-game user interaction
information collection and analysis
6. T-MON: Monitor of xAPI-SG traces
Experience API Profile
for Serious Games
(xAPI-SG)
T-MON
Monitor of xAPI-SG traces
Jupyter Notebooks
Default set of analysis
and visualizations
7. T-MON: Monitor of xAPI-SG traces
Monitoring xAPI-SG traces:
1. Select local or remote mode
2. Select xAPI-SG data file
3. Run the analysis!
It allows to configure the data:
- filter players
- filter game parameters
and configure the visualizations:
- order data
- view percentage/total
- plot only partial data per graphic
8. Default analysis and visualizations (i)
Serious game completion
initialized and completed
traces with object-type
serious-game
Choices in alternatives
selected traces with
object-type alternative,
result.response and
result.success
Serious game progress
initialized, progressed and
completed traces with object-type
serious-game, result.progress and
timestamp
Completable progress
progressed traces in any
completable object type, with
result.progress
9. Default analysis and visualizations (ii)
Completable results
(scores)
completed trace of any
completable with result.score
Completable results
(times)
difference in timestamp
of initialized and
completed traces of
each completable
Interactions
interacted traces with any
object type; bar chart per item,
and each bar per player
Interactions (heatmap)
interacted traces grouped by
item (object) and player
and more!
10. T-MON: Monitor of xAPI-SG traces
T-MON: monitor of xAPI-SG traces, open-source
code available at:
https://github.com/e-ucm/t-mon
● Local mode to run Notebooks in your local-hosted
Jupyter server
● Remote mode to run Notebooks in a web-hosted Jupyter
server
○ Including launch with binder from GitHub
no installation or configuration required!
11. Conclusions
➔ Standard xAPI-SG covers most common interactions with SGs
➔ Default set of analysis and visualizations in T-MON using xAPI-SG,
with no configuration required
■ For data scientists, entry-point to game information,
without game-specific or xAPI-SG knowledge
■ For game designers, ready-to-use analyses,
without extensive data analysis expertise
➔ Simplify the report of the collected interaction data from SGs remotely
with the online tool (e.g., New Normal contexts)
12. Future work
➔ Geolocalized games
■ Extension of the xAPI-SG Profile
■ Games created with uAdventure authoring tool
■ Location-based analyses and visualizations in T-MON
■ Particularly adequate in the New Normal
➔ Extend the analyses and visualization of T-MON with other
xAPI-compliant data (beyond xAPI-SG)
➔ Integration with SIMVA to validate games and collect analytics
www.e-ucm.es/uadventure
www.e-ucm.es/simva