Visual 1
Several weeks ago we discussed ways how to determine which teams are best in current game statistics. Mainly we want to determine which team has the best defense and offense. For such case we are using plots made in Vega and later improved using holoviz and panel. Unfortunately we still have problem with sharing. We tried the tutorial on a course webpage and we managed to run it on a server, but we still did not figure out how to export it to html to publish it on a blog. We would be really happy for any advice. Because we are unable to share the visual directly we decided to put here a link for the jupyter notebook the visual is created at: link
The visuals work if they are run in Jupyter, but if we export notebook to html the is a blank page only with widgets instead of visual, so we were unable to share it here on the blog. We are really sorry for inconvenience.
We decided to share at least screenshots of visual which can be run in the notebook. There are three tabs which can be used to determine best teams. In the first picture one can choose conference and division of teams and decide he/she wants to know about defense or offense of the teams.
Picture with defensive stats in the south-eastern division.
Picture with offensive stats in the pacific division.
In the second and third tab specific teams can be chosen by their names to compare teams from different divisions. In second tab there are offensive statistics and in the third tab there are deffensive ones.
Offensive statistics by team.
Defensive statistics by team.
We think that these visuals can be used to determine which teams are the best in particular parts of the game. To the final project we will think also about versions filtered by a season or versions containing data only from matches from teams with different conferences (to determine which conference is better).
Visual 2
One of our essential research questions was to try to divide the teams into two groups, defensive and offensive teams, based on how they tend to play. In order to make this division, multiple different statistics, from offensive and defensive point of view needs to be analyzed, and the following two visuals were built to answer that need. Both visuals have four scatter plots, upper one with statistics related to teams' offensive abilities, and the latter one with stats related to the defensive abilities of the teams. Both visuals allow brushing, so that the performance of certain teams can be highlighted in all of the four plots. Both visuals also have a widget, from where the user can choose to look at the statistics of a specific season, or the average statistics across all seasons. The visuals were implemented as vega-lite specifications, and they can be found as github gists from here (offensive) and here (defensive).
Offensive statistics:
Defensive statistics:
However, as no clear clustering to defensive and offensive teams can easily be found using just these statistics, different statistics and pairwise comparisons still needs to be experimented. In addition, a widget with which the user can choose which teams to include in the visuals will be implemented to make comparison between specific teams easier.
_______________________________________________________________________
Visual 3
From previous post, we tried to visualize and answer the question regarding whether fair play may have an affect on the Total_points a team scored, which then relates to winning tendency of the team. The Winning or Losing result is marked by the colors of the points. While we expected a fair-play team should have more chance to win, surprisingly, the Total_Fouls factor does not contribute much on the wining tendency.
We then try to expand the visual by adding the selection so that the viewers can choose to see the result in each season. The codes can be found through this link.