23
SCIENCE * PASSION * TECHNOLOGY HOW PLAYSTYLES EVOLVE: PROGRESSION ANALYSIS AND PROFILING IN JUST CAUSE 2 JOHANNA PIRKER, TU GRAZ, AUSTRIA SIMONE GRIESMAYR, TU GRAZ, AUSTRIA ANDERS DRACHEN, AALBORG UNIVERSITY & THE PAGONIS NETWORK, DENMARK RAFET SIFA, FRAUNHOFER IAIS, GERMANY SEPT-28:: IFIP ICEC2016, VIENNA

How Playstyles Evolve: Progression Analysis and Profiling in Just Cause 2

Embed Size (px)

Citation preview

S C I E N C E * PA S S I O N * T E C H N O L O G Y

HOW PLAYSTYLES EVOLVE: PROGRESSION ANALYSIS AND PROFILING IN JUST CAUSE 2

J O H A N N A P I R K E R , T U G R A Z , A U S T R I A S I M O N E G R I E S M AY R , T U G R A Z , A U S T R I A

A N D E R S D R A C H E N , A A L B O R G U N I V E R S I T Y & T H E PA G O N I S N E T W O R K , D E N M A R K

R A F E T S I FA , F R A U N H O F E R I A I S , G E R M A N Y

S E P T- 2 8 : : I F I P I C E C 2 0 1 6 , V I E N N A

Design

AnalysisDesign

Games Mechanics

Experiences

Immersion

AudioAnimation

Graphics / Objects

Character (1st / 3rd)

Interactivity

Interface

Challenges Quests, Puzzles,…

BARTLE’S GAMER TYPES

http://www.gamerdna.com/quizzes/bartle-test-of-gamer-psychology

FLOW EXPERIENCE

http://www.gamerdna.com/quizzes/bartle-test-of-gamer-psychology

GAME ANALYTICS

▸ Understanding player behaviour to create better game experiences

▸ Understanding and identifying patterns in player data

▸ -> who is the player?

▸ -> statistics on player behaviour (retention rate, concurrency, )

▸ …

Further reading: El-Nasr, M. S., Drachen, A., & Canossa, A. (2013). Game analytics: Maximizing the value of player data. Springer Science & Business Media.

BEHAVIOURAL PROFILING::CLUSTER ANALYSIS

▸ Finding patterns in behavioural game data

▸ Unsupervised learning strategies to find groups/clusters of players playing in a similar way / fit various patterns

▸ identify groups with similar behaviour and identify the most important behavioural features in terms of underlying patterns in the dataset

Further reading: http://blog.gameanalytics.com/blog/introducing-clustering-behavioral-profiling-game-analytics.html

PROGRESSION ANALYSIS AND PROFILING IN JUST CAUSE 2

MAIN CONTRIBUTION

▸ Behavioural profiling through clustering with Archetypal Analysis (AA) combined with progression analysis in an Open-World game

▸ The main storyline of Just Cause 2 to measure progression along multiple vectors

▸ Sankey flow diagram for a visual inspection

JUST CAUSE 2

▸ Progression along different vectors, seven Agency-related missions, missions from a number of Rebel Factions, Stronghold missions

▸ All mechanics in game available from the beginning (direct gameplay approach)

DATASET

▸ Dataset provided by Square Enix ▸ Play histories from over 5000 JC2 players (2010) ▸ Various behavioural features collected: ▸ actions with ▸ in-game geographical coordinates ▸ timestamps

▸ metrics from the gameplay ▸ e.g. total kills, total chaos, kilometres driven # of

stronghold takeovers ,…

▸ Data set pre-processing (cleaning): ▸ Outliers removed: scores outside 1-99th percentile

excluded ▸ (faulty tracking or errors)

FEATURES

▸ Agency missions (+ reach specific level of Chaos)

▸ subset of features based on the core mechanics

▸ -> does not impact the analytical framework

▸ -> impacts the kinds of conclusions that can be derived

ANALYSIS & RESULTS

FEATURES

▸ Spatio-temporal navigation

▸ combat performance

▸ progression through the main storyline

▸ side quests..

▸ Agency missions (+ reach specific level of Chaos)

▸ subset of features based on the core mechanics

▸ -> does not impact the analytical framework

▸ -> impacts the kinds of conclusions that can be derived

PLAYER PROGRESSION ALONG THE MISSIONS

ANALYSIS

▸ Archetypal Analysis (AA) for behavioural profiling

▸ AA models applied to all seven agency mission bins

▸ Optimal # of clusters (k) determined for each (analysis of the residual sum of squares for all k value less than or equal to 20, and chose the number of clusters with the elbow criterion)

▸ -> three main archetypes

PLAYER PROFILES

PLAYER BEHAVIOUR ALONG THE STORYLINE

RESULTS

▸ How does in-game behaviour and performance change over the various missions?

▸ (see Sankey diagram)

▸ player behaviour changes - players do not remain in a single cluster (also due to the nature of the mission design)

▸ domination in exploration-based features (e.g. playtime)

RESULTS

▸ How many profiles enter players on average over the course of the game?

▸ They change at least once

▸ Avg. 2.91 clusters

RESULTS

▸ How can we describe player behaviour of the different player profiles?

GOALS

• Improve our understanding of the different player behaviours and factors to improve engagement

• Find issues to avoid drop-outs

• Provide tools for game designers to (visually) analyse the game and improve the understanding of players

• Find game design flaws early and maybe also automatically/dynamically

THANK YOU FOR YOUR ATTENTION.

JOHANNA PIRKER, [email protected], @JOEYPRINK

Further information: andersdrachen.com jpirker.com

Thanks to Simone, Anders, and Rafet!! Thanks to Square Enix! Thanks to the reviewers!