现在我需要创建一个自己的数据挖掘任务。我已经和一些人交谈过，最流行的想法是价格预测或体育结果预测，我认为已经有很多人在实现它们。
那么，有人可以给我一些现实生活中的想法吗？您发现数据挖掘可能有用，例如根据顾客在超市已经购买的商品来预测顾客想要购买的商品。
任何想法都会受到欢迎，提前致谢。
Now I need to create a data-mining task of my own.I already talked to some people,the most popular ideas would be price prediction or sport result prediction,which I think there are already plenty of people implementing them.
So could anyone give me some real-life ideas please that you found data-mining may be of use,like predicting what the customer would like to buy based on what they already purchased in a supermarket.
Any idea would be welcomed,thanks in advance.
发布评论
评论（3）
我发现数据挖掘在网站上进行链接点击分析时很有用。该信息用于通过微调网站最常用的导航来优化用户体验。
它与非常具体的链接和链接的“类别”相关，例如“小项目描述”和“大项目标题”
I found data mining useful when conducting link-click analysis on websites. This information was used to optimize the user-experience by fine-tuning the most appriciated navigation through the site.
It related to very-specific links and 'classes' of links such as "small item description" and "large item title"
找到可以轻松解析的数据源是关键，那里有很多很酷的数据集，请查看 statlib 对于一些。目前更流行的数据挖掘应用程序之一是可视化。您可以采用一个大型（但不是太大）数据集，其中每个元素以某种方式彼此相关，并使用降维算法对其进行可视化。多维缩放 (MDS)、ISOMap、最大方差展开 (MVU) 和结构保留嵌入等算法可以创建数据的 2D（或 3D）表示。
Finding a source of data that can be easily parsed is key, there are quite a few cool data sets out there, check out statlib for a few. One of the more popular data mining applications these days is visualization. You can take a large (but not too large) dataset where each element is related to each other in some way and use dimensionality reduction algorithms to visualize it. Algorithms like multi-dimensional scaling (MDS), ISOMap, maximum variance unfolding (MVU), and structure preserving embedding can create 2D (or 3D) representations of the data.
http://en.wikipedia.org/wiki/Machine_learning#Applications
http://en.wikipedia.org/wiki/Machine_learning#Applications