Heatmaps are an excellent tool for visualizing data with a two dimensional spatial component. They are frequently used to map out player deaths in shooter games. The most common heatmaps use the location of the victim and aggregate the number of kills as the visualized measure. However, this simple metric doesn’t tell the whole story. With a few simple tweaks we can build balance and flow maps, which can be useful tools for understanding the way players use the map.

Studies have shown that a good review score can propel a game’s reputation. However, with such a large amount of data available, the two review aggregation sites simply present a single number for each game. With a wealth of review score data available, a few web scraping scripts, and a couple data analysis tools, some very interesting patterns emerge. The Data The following charts are backed by 46,000 review scores for 800 games.

In Part I we looked at review scores in isolation. When we combine review score data with sales data some interesting patterns emerge. This section includes sales data from around 7,000 Xbox360 and Playstation2 games as well as the review score data from Part I Sales and Scores An interesting property of the sales data is the lack of sales variation below a score of 70. Games that score above this threshold show a clear trend of increasing sales.

Scraping Data All of the data used to build these visualizations was extracted from various web sites using a few Python scripts. The Beautiful Soup library was used to parse and traverse the HTML. Google Chrome’s developer tools are an excellent way to discover the structure of a web site so it can be traversed quickly with scripts. Scraping data from web pages is easy as long as you completely ignore the impulse to follow good coding practices.