有什么简单的性能数据探索程序吗?
我正在尝试优化一些软件,因此我生成了大量现实世界的性能测量结果 - 没什么花哨的,只是描述情况的几个数字加上以毫秒为单位的时间。
我对其进行了一些基本分析 - 主要是以各种方式将数据划分为桶并计算桶平均值 - 这对于让我有一个总体想法非常有帮助,但似乎这些关系比我预期的更复杂。
与此同时,我将继续对数据使用各种公式,但也许有一个机会,我可以使用一个工具来直观地探索这些数据,并以这种方式寻找模式?有什么建议吗?
I'm trying to optimize some software, so I generated a large volume of real world performance measurements - nothing fancy, just a few numbers describing the case plus time in milliseconds.
I did some basic analysis on it - mostly dividing data into buckets in various ways and calculating bucket averages - and it was quite helpful in giving me a general idea, but it seems these relationships are more complex than I expected.
In the mean time I'll keep throwing various formulas at the data, but perhaps by some chance there exists a tool I could use to explore such data visually, and look for patterns this way? Any recommendations?
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(2)
如果您愿意花一些钱,Tableau 和 Spotfire 擅长可视化几乎任何类型的数据。
If you are willing to spend some money, Tableau and Spotfire are good at visualizing data of practically any kind.
我喜欢使用 Excel 进行此类原始数据性能分析。将原始数据转储到 .csv 文件中,将其加载到 Excel 中,然后您可以根据需要对数据进行分组和绘制图表。一旦绘制成图表,通常就会出现可辨别的模式。
I like Excel for this sort of raw data performance analysis. Dump your raw data into a .csv file, load it up in Excel and from there you can group and graph the data however you want. Once graphed, often discernible patterns will emerge.