For easier reading, I left out all possible buckets and focused on interpreting the change_points_count. A significant trend change was detected on the March 12 — a day after GothamChess released the video, which makes sense as he lives in New York City and, therefore, the time zone impacts this. The detected type is a trend_change. Like before, the p-value is associated with a question in our candy example. The change point aggregation asks many questions, like is this a spike, step change, distribution change, and many more. Furthermore, the answer gives the change_point, which is 77. This relates to the 77t bucket of the date_histogram. Then we have the p_value, a tiny number `2.6494782968908805e-38` equals to this: 0.000000000000000000000000000000000000026494782968908805. Statistically, significance is typically mentioned when the p-value is 0.05 or 0.005. Our p-value is much lower than that. Therefore, we can safely assume that the impact GothamChess has is significant.
Running the same query with London instead of Ponziani to analyze the significance of Hikaru Nakamura’s video did not detect a significant difference.
Summary
This blog post investigated streamers’ and YouTubers’ influence in chess. We showed how one could calculate p-values directly in the Elastic Stack. This lighthearted introduction to p-values showcases how you can use them to detect significant changes and that it does not always need to be for most of your data.
Change point detection is a way to surface things that are difficult or nearly impossible to eyeball from a chart — but this analysis has some caveats. We only checked the impact of a single video from GothamChess. It could be that different chess streamers, such as Alexandra Botez (Twitch: 1.2 million followers, YouTube: 1.27 million subscribers)* played a few Ponzianis on stream. We did not verify if any tournaments were happening where the Ponziani was played. Additionally, we did not analyze the distribution of the rating of Ponziani players. Furthermore, we only verified data on Lichess. Likewise, this analysis only includes rated games, where players can lose/gain ratings. Now I am off to play some Ponziani Openings!
Ready to get started with Elastic? Begin a free 14-day trial of Elastic Cloud. Or download the self-managed version of the Elastic Stack for free.
* All values were retrieved on April 4, 2023, from the respective YouTube channel and Twitch.
Leave a Reply