Cody问题可视化，重新审视

Posted byNed Gulley，，，，2020年8月7日

4views (last 30 days) | 0Likes | 0comment

Way back in 2013 I did ablog post about visualizing Cody problems。The idea was to take various metrics for each problem and let people plot them in different ways. The结果可视化仍在运行，，，，but the data set only includes problems that were created before February 2013. I’ve wanted to revisit this visualization for some time. 2013 was a long time ago! These days, instead of 1200 problems to look at, we have more than 3700.

This summer, with the help of my hard-working intern Anirudh Watturkar, we got the old visualization up and running again using the latest data. Here it is:

科迪Problem Visualization

只需使用下拉菜单选择X轴和Y轴即可。您还可以使用两个具有点尺寸和颜色的附加尺寸。

而且，一旦您拥有喜欢的情节，就可以悬停在一个点上，以查看它与之相对的问题，也可以单击它并将其运输到该问题。

马上，您会注意到降落时的默认视图，ID与ID，看起来很奇怪。这是因为我们在某一时刻有大量垃圾邮件问题的大量涌入，我们消除了30,000个虚假问题。因此，ID在2016年中期在某个地方进行了戏剧性的跳跃。我添加了一个名为Quards_Index的维度，该维度从1到最近的问题依次计数，丢弃了所有垃圾邮件条目。

如果我们查看问题年龄与问题索引，我们可以看到问题的速度更快，更慢。水平的“高原”形状意味着在短时间内出现了许多问题。这些通常发生在比赛和晋升期间。

如果我们查看furety_id vs.年龄，您可以看到水平模式，这些模式向那些创造了许多问题的作者显示。水平乐队越长，该人作为作者的活跃时间就越长。看到作者的贡献涵盖了很多年，真是令人印象深刻。但是，当一个人很快出现问题时，这也令人印象深刻。

我们看看author_id与percent_correct。你can see that, of those authors with many problems, some skew toward being harder (low percent_correct) while others skew toward the easy (high percent_correct). A small set of problems are pegged against the 100% correct wall.