Comments on: How to: make a scatterplot with a smooth fitted line http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/ Strength in Numbers Thu, 24 May 2012 15:48:10 +0000 hourly 1 http://wordpress.org/?v=3.3.2 By: Nathan Yau http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-43558 Nathan Yau Fri, 07 May 2010 14:17:44 +0000 http://flowingdata.com/?p=6479#comment-43558 instead of scatter.smooth, use loess.smooth (with the same variables). that'll show you the x and y values of the curve. Like this: p <- loess.smooth(x=1:length(unemployment$Value), y=unemployment$Value) So if you type 'p$x' that'll give you the x-values. After that, read up on write.table() to write to a text file: http://pbil.univ-lyon1.fr/library/base/html/write.table.html instead of scatter.smooth, use loess.smooth (with the same variables). that’ll show you the x and y values of the curve. Like this:

p < - loess.smooth(x=1:length(unemployment$Value), y=unemployment$Value)

So if you type 'p$x' that'll give you the x-values.

After that, read up on write.table() to write to a text file:

http://pbil.univ-lyon1.fr/libr.....table.html

]]> By: Andrea http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-43556 Andrea Fri, 07 May 2010 14:06:02 +0000 http://flowingdata.com/?p=6479#comment-43556 This is probably the wrong place to post this question but how do I write the values of the fitted curve into a text file? Thanks, This is probably the wrong place to post this question but how do I write the values of the fitted curve into a text file?
Thanks,

]]> By: Notional Slurry » links for 2010-03-30 http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-41563 Notional Slurry » links for 2010-03-30 Wed, 31 Mar 2010 06:05:23 +0000 http://flowingdata.com/?p=6479#comment-41563 [...] How to: make a scatterplot with a smooth fitted line | FlowingData "Oftentimes, you'll want to fit a line to a bunch of data points to make it easier to spot patterns or relationships. It might be observations over time or it might be two variables that are possibly related. In either case, a scatter plot just might not be enough to see anything useful. This tutorial will show you how to graph a fitted line, or loess curve, to such a scatter plot." (tags: tutorial visualization graphs R statistics how-to) [...] [...] How to: make a scatterplot with a smooth fitted line | FlowingData "Oftentimes, you'll want to fit a line to a bunch of data points to make it easier to spot patterns or relationships. It might be observations over time or it might be two variables that are possibly related. In either case, a scatter plot just might not be enough to see anything useful. This tutorial will show you how to graph a fitted line, or loess curve, to such a scatter plot." (tags: tutorial visualization graphs R statistics how-to) [...]

]]> By: Data Visualization Resources - Combining and Filtering Time Series and Static Data Sets in EX Dashboards http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-41555 Data Visualization Resources - Combining and Filtering Time Series and Static Data Sets in EX Dashboards Wed, 31 Mar 2010 01:06:18 +0000 http://flowingdata.com/?p=6479#comment-41555 [...] How to: make a scatterplot with a smooth fitted line | FlowingData [...] [...] How to: make a scatterplot with a smooth fitted line | FlowingData [...]

]]> By: Jon Peltier http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-41550 Jon Peltier Tue, 30 Mar 2010 22:41:32 +0000 http://flowingdata.com/?p=6479#comment-41550 Kelly - I agree. Using the algorithm based on the NIST documentation, I could not duplicate Nathan's smoothing. With a large enough span (>=66%) my fit doesn't curve upwards and its general shape is similar, but it doesn't sink as low as Nathan's does at the end. Kelly -

I agree. Using the algorithm based on the NIST documentation, I could not duplicate Nathan’s smoothing. With a large enough span (>=66%) my fit doesn’t curve upwards and its general shape is similar, but it doesn’t sink as low as Nathan’s does at the end.

]]> By: Mark James Adams http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-41546 Mark James Adams Tue, 30 Mar 2010 21:59:27 +0000 http://flowingdata.com/?p=6479#comment-41546 The exact line you get will be sensitive to the parameters you feed into loess (the size of the local regression window). If nothing else, though, Nathan's chart accentuates how out of wack the decrease in employment is since 2008—way more than structural. The exact line you get will be sensitive to the parameters you feed into loess (the size of the local regression window).

If nothing else, though, Nathan’s chart accentuates how out of wack the decrease in employment is since 2008—way more than structural.

]]> By: Tom http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-41544 Tom Tue, 30 Mar 2010 21:52:45 +0000 http://flowingdata.com/?p=6479#comment-41544 Or you could use Rapidminer 5.0 and their time series operators. You could build a nonlinear trend line using Support Vector Machines or a standard Neural Net for this time series data. Or you could use Rapidminer 5.0 and their time series operators. You could build a nonlinear trend line using Support Vector Machines or a standard Neural Net for this time series data.

]]> By: Kelly O'Day http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-41539 Kelly O'Day Tue, 30 Mar 2010 20:27:17 +0000 http://flowingdata.com/?p=6479#comment-41539 I have a problem with the trend line. Looking at the monthly values, I notice a substantial rise after 2008, consistent with news reports of our economy. Your trend line does not respond to the increase in unemployment starting around 2008, rather it continues a decline. I have reproduced your chart, however, my loess trend line shows the appropriate increase after 2008. Not sure why the discrepancy. I have a problem with the trend line. Looking at the monthly values, I notice a substantial rise after 2008, consistent with news reports of our economy.

Your trend line does not respond to the increase in unemployment starting around 2008, rather it continues a decline.

I have reproduced your chart, however, my loess trend line shows the appropriate increase after 2008.

Not sure why the discrepancy.

]]> By: Mark James Adams http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-41510 Mark James Adams Tue, 30 Mar 2010 08:40:50 +0000 http://flowingdata.com/?p=6479#comment-41510 A nice post would one that starts a discussion about when you need to break into Illustrator/Inkscape to make your point. I'm with John that you can get pretty far in R for making your graph publication ready, so when do you need that extra bit of polish that a drawing tools will provide? (didn't Tufte say something about needing one tool that can count and one tools that can see). A nice post would one that starts a discussion about when you need to break into Illustrator/Inkscape to make your point. I’m with John that you can get pretty far in R for making your graph publication ready, so when do you need that extra bit of polish that a drawing tools will provide? (didn’t Tufte say something about needing one tool that can count and one tools that can see).

]]> By: John Christie http://flowingdata.com/2010/03/29/how-to-make-a-scatterplot-with-a-smooth-fitted-line/#comment-41507 John Christie Tue, 30 Mar 2010 03:27:33 +0000 http://flowingdata.com/?p=6479#comment-41507 thanks... that second one was what I had planned to paste but I picked up the wrong thing thanks… that second one was what I had planned to paste but I picked up the wrong thing

]]>