Re: [R] selecting outliers

From: Søren Højsgaard <>
Date: Mon 08 Aug 2005 - 22:59:35 EST

Perhaps what Alessandro is after is simpler than that: Making a plot of data in a data frame, being able to click on 'suspicious points', getting the corresponding rows of a data out in a new data frame (for further inspection) while keeping the 'good points' in the plot (and perhaps redoing some calculations on the basis of the good points only....). This could then go on in an iterative way. That would be a perfectly sensible thing to do. How difficult it is technically I don't know, but it seems that it would require a call-back mechanism from a plot window to R (and a more 'advanced' one than provided by 'locator()').

Best regards

-----Oprindelig meddelelse-----
Fra: [] På vegne af Christian Hennig Sendt: 8. august 2005 14:45
Til: alessandro carletti
Emne: Re: [R] selecting outliers

Hi Alessandro,

On Mon, 8 Aug 2005, alessandro carletti wrote:

> Hi everybody,
> I'd like to know if there's an easy way for extracting outliers record
> from a dataset, in order to perform further analysis on them.

The answer is "no". The reasons are not technical. There are some quite easy outlier detection approaches around (e.g., compute robust Mahalanobis distances with and call the points with too large distances "outliers"). But the main problem is that the term outlier has no objective, unique meaning. It depends crucially on your aims and on the assumptions you want to make about the non-outliers in the dataset (which should be elliptically distributed and homogeneously close to a multivariate normal distribution for the Mahalanobis approach).

Christian mailing list PLEASE do read the posting guide! mailing list PLEASE do read the posting guide! Received on Mon Aug 08 23:02:53 2005

This archive was generated by hypermail 2.1.8 : Fri 03 Mar 2006 - 03:39:45 EST