Re: Unbearably slow plotting (v6)
- To: mathgroup at smc.vnet.net
- Subject: [mg79652] Re: Unbearably slow plotting (v6)
- From: Bill Rowe <readnewsciv at sbcglobal.net>
- Date: Wed, 1 Aug 2007 05:00:08 -0400 (EDT)
On 7/31/07 at 6:00 AM, j.f.b.payne at tesco.net wrote: >Bill Rowe wrote >>With 1E5 and 1E6 points, this results in a plot that is >>indistinguishable from a filled rectangle. That seems to be of very >>little use. >Well, yes, in this _example_ >But one might have a dataset with 99.99% of the data on a line, with >some outliers. The 99.99% data will be overplotted (many points on >same pixel) but the outliers will show up in the plot. Not necessarily. More often than not, the values won't be precisely the same and there will be a cluster of pixels representing a cluster of points which tends to obscure the trend in the data. And with default settings for ListPlot, outliers are often suppressed. >And, one might not want to spend time writing and testing code to find >the outliers (code which, in any case, is liable to have a run time as >bad as ListPlot). Agreed. But there are existing functions in the standard distribution of Mathematica that are better designed for detecting outliers than simply using ListPlot. For example, in the package StatisticalPlots there is the function BoxWhiskerPlot which is produces a far better plot for finding outliers than ListPlot. -- To reply via email subtract one hundred and four