INSPECT

Visualization Tool Generating Insights
on Process Data

Introduction

Player journey maps visualize all player actions in chronological order. Such visualization can help researchers reveal the sequential relationship of player actions and understand their play experiences. This also generalizes to customer journey maps for websites and mobile applications. Visualizations of journey maps can help analyze player behaviors, player strategies, as well as allow designers to validate their own assumptions of player experience. One critical challenge for such visualizations is the clarity and readability of the visualized graphs. If one visualizes all players' actions in a granular manner, then graphs can become cluttered. INSPECT is an interactive visualization tool that was developed to overcome this problem. It gives designers the ability to generate player journey maps. Moreover, it includes several features that can help simplify graphs and segment players to allow designers to focus on specific player groups. See below for more information and demonstrations.

INSPECT's FEATURES AND APPLICATIONS

Below we show some projects. Here we will show what INSPECT's features are.

INSPECT Presentation

Projects

Player segmentation with INSPECT

In this project, we visualized player data from two games: Guild Wars 2 and Parallel to analyze player behavior. For player data from Guild Wars 2, we segment players based on their experience level. By comparing experience and inexperience player journeys, the visualizations can show clear differences in the decision-making strategies. For player data from Parallel, we segment players based on their behaviors, which allowed us to detect desired and undesired behaviors.

Combo Analysis with INSPECT

In this project, we visualized player data from Guild Wars 2 to understand how players use their skills. For all sequences shorter than 10, we calculate the probability of each sequence based on its occurrence to the total occurrence of all sequences with the same length. Then we calculate the average probability for each length from 2 to 10, deploying a scatter plot to visualize the difference between each sequence of length n and the average probability of all sequences in length n. By ranking all sequence of length n (n<10) and their distance to the average probability, we are able to understand which sequence is more used in the game.

Explainability Analysis with INSPECT (on-going)

In this project, we visualized player data from Wake to understand how players go through their job sequences.

Smart City Analysis with INSPECT

Given this visualization, we applied it to a different dataset. A dataset on smart cities indicators. In this project, we visualize smart city data from the report, https://www.imd.org/smart-city-observatory/home/. We developed two kinds of visualizations: correlation graphs and performance graphs. Visualizing correlations as graphs rather than tables shows various relationships between variables and can allow us to see clearly the third variable effect. The video below shows the correlation results we got for the smart city data. We further used the graph representation in INSPECT to develop performance graphs, where nodes represent cities and indicators and edges represent performance. By comparing the performance graphs of four cities (top 2 and bottom 2) in Africa and Europe, we can deduce differences between cities as discussed in the video below.