寻找集群的东西 [英] looking for something to cluster

问题描述

大家好。

我今年仍然在为毕业设计创意

我决定在c＃中实施3种clustring算法来比较它们的性能cllicking但仍然不知道要集群什么

我想要类似Fisher Iris（同样的难度级别），因为项目的集中点是算法的比较



感谢所有

Hello everyone
I'm still working on an idea for graduation project this year
I've decided to implement 3 clustring algorithms in c# to compare their performance in clustring but still don't know what to cluster
I want something like Fisher Iris (the same difficulty level)because the concentration point of the project is in the comparison of the algorithms

Thanks to all

推荐答案

为什么不使用Fisher Iris数据集？我要问的问题是，这个相当简单的数据集是否真的会运用你所比较的算法，从而产生相对性能和功能。



如果我是评估你的研究的学术人员（不幸的是你！），我会看你做的定性研究来选择类型的数据集你工作过的非常重要。



我希望你通过熟练掌握你测试的算法的历史来保护你的选择，关于他们最熟悉的算法 - 以及最糟糕的性能案例，并对您选择的数据集如何有助于比较群集性能提供可靠的解释。



而且，我会问你问题喜欢：给出一个非常大的数据集来分析许多可能的因素：按成本描述您的算法选择（

Why not use the Fisher Iris dataset ? The question I would ask would be if that rather simple dataset really would "exercise" the algorithms you compare in ways that brought out their relative performance, and capabilities.

If I were a member of the academic staff evaluating your research (unlucky you !), I would be looking at the qualitative research you did to select the types of dataset(s) you worked with as being extremely important.

I would expect you to "defend" your choices by speaking knowledgeably about the history of the algorithms you test, about their known best- and worst- performance cases, and giving a solid explanation of how the datasets you chose are useful to compare clustering performance.

And, I'd be asking you questions like: "given a very large dataset to analyze with many possible factors: describe your choice of algorithms by cost(

，CPU时间），以及对硬件资源的需求（存储器） y，数据存储，带宽）。



你可以在网上找到各种类型的各种数据集，只需要一点点搜索。



这是一堆：[ ^ ]。

, CPU time), and demands for hardware resources (memory, data storage, bandwidth)."

You can find all kinds of datasets of every type on the web with just a little searching.

Here's a bunch: [^].

这篇关于寻找集群的东西的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！