tag:blogger.com,1999:blog-6597942605529665284.post8021267230168784980..comments2024-03-08T00:38:36.448-08:00Comments on Machine Learning with JSAT: Probability Estimation: Kernel Density EstimatorsAnonymoushttp://www.blogger.com/profile/04050596766362107081noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-6597942605529665284.post-72642740114847520652016-06-08T12:53:40.190-07:002016-06-08T12:53:40.190-07:00Hi Edward,
I'm, so glad to have found your bl...Hi Edward,<br /><br />I'm, so glad to have found your blog. You are the first one being able to explain ML without writing dozens of equations in the first paragraph. Thanks for that.<br /><br />Now I'm getting close to the solution of my problem. I want to determine the value of comic books based on some characteristic (say number of prints, year of publication, physical condition). As far as I understand I could use KDE as follows:<br />- create tuples (char1, char2, char3, value) of my training set<br />- train algorithm, i.e. MetricKDE<br />- To get the value of a new comic having char1, char2, char3 I would have to estimate the value several time, until I find the tuple with the highest pdf.<br /><br />Is that correct? Does there already exist an algorithm finding the value with highest pdf?<br /><br />Regards,<br />MichaelAnonymoushttps://www.blogger.com/profile/15837654229717617734noreply@blogger.com