特征是否可以成为聚类算法的列表？
这是我的用例：
我有网络资源，并且每个资源都有一组视图。它来自网络导航，因此资源是您访问的内容，浏览量是实际访问的内容。
对于每个视图，我都有有关位置、创建视图的时间的信息，并且我可以获得诸如视图总数和在该资源上传递的时间之类的信息。
我想在这些信息中找到不同的模式，以确定用户配置文件。如果对于每个资源我只有一个视图，那么将它们集群起来可能很容易，但是由于我确实有许多视图，有没有一种方法可以使用这些信息来更好地集群资源？
Here is my usecase:
I have web resources and for each resource a set of views of it. It's coming from web navigation, so resources are what you visit and views are actual visits.
For each view, I have information about the location, the time of the day the view was made, and I can get information like the total number of views and the time passed on this resource.
I want to find different patterns in this information, to determine user profiles. If, for each resource I had only one view, it could be easy to cluster them, but as I do have a number of views, is there a way I can use this information to better cluster resources?
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）
如果我理解正确的话，您希望根据资源拥有的视图对资源进行集群。为此，您可能需要根据视图的统计数据创建一个特征向量。例如，您的特征向量可以包括：观看总数、观看者位置的标准差、观看的平均时间、资源的年龄等。然后，您可以将这些特征向量输入到聚类算法中。您想要使用的确切功能取决于您对集群的使用。如果您想让集群反映资源的内容并且资源是基于文本的，那么我建议使用 bag单词也是如此。
If I understand correctly, you wish to cluster the resources based on the views they have. To do this, you may want to make a feature vector based on the statistics of the views. For example, your feature vector can include: total number of views, std dev of the locations of viewers, mean time of day of views, age of resource, etc. You can then feed these feature vector into a clustering algorithm. The exact features you want to use will depend and your use of the clusters. If you want to have cluster reflect the content of resources and the resources are text based, then I would recommend using a bag of words as well.
很难说出你到底在问什么。是的，特征向量广泛应用于机器学习算法中。
Hard to tell what exactly you are asking. Yes, feature vectors are widely used in machine learning algorithms.