Dear Rich, Assuming that I understand what you want to do, try adding the following to your script (which, by the way, is more complicated that it needs to be): xx <- 10:50 m <- lm(y ~ x) yy <- predict(m, data.frame(x=xx)) lines(spline(xx, yy), col="blue") m <- lm(y ~ log(x)) yy <- predict(m, data.frame(x=xx)) points(xx, yy, col="magenta") The first set of commands adds a line corresponding to the points that you plotted, which if I understand right, is *not* what you want. The second set of commands shows how to find points along the diagonal straight line that you plotted, given their x-values, which is what I think you want. If you examine the linear models fit, you'll see that they just interpolate between the two points, albeit differently. I hope this helps, John ----------------------------- John Fox, Professor Emeritus McMaster University Hamilton, Ontario, Canada Web: socserv.mcmaster.ca/jfox> -----Original Message----- > From: R-help [mailto:r-help-bounces at r-project.org] On Behalf Of Evans, > Richard K. (GRC-H000) > Sent: Monday, September 25, 2017 3:28 PM > To: r-help at r-project.org > Subject: [R] bowed linear approximations > > Hello, > > Please run the following code snippet and note the resulting plot: > > x <- c(10, 50) > y <- c(0.9444483, 0.7680123) > plot(x,y,type="b",log="x") > for(i in 1:50){ > xx <- exp(runif(1,log(min(x)),log(max(x)) )) yy <- approx(x,y,xout=xx, method > "linear") > points(xx,yy$y) > } > > notice the "log=x" plot parameter and the resulting "bow" in the linear > approximation. > > This makes sense when I realized that the plot command is first making the > plot and then drawing straight lines between points on a log plot AFTER the > plot is generated and that that's why the line is straight. I get that. > .. and it also makes sense that the bowed points are a result of the linear > approximations being made BEFORE plotted in a logarithmic plot. I get that.. > > My goal is to make approximations that lie on the line produced on the plot as > shown, so I realize that what I want to do is NOT linear approximations, but > maybe "log" approximations? > However, the approximation methods are only "linear" and "constant" .. there > isn't a "log" method to approximate with. > > So can anyone tell me how to fix the code such that he approximated points > DO lie on the line as plotted with the "log=x" plot parameter? > Oh, and they have to be uniformly distributed along the Log=x axis.. I think > that's the tricky part. > > Any help and/or insight would be greatly appreciated. > > Thank you! > -Rich > > > [[alternative HTML version deleted]] > > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting- > guide.html > and provide commented, minimal, self-contained, reproducible code.
Hi Rich, If I understand your comment about "uniformly distributed along the log=x axis" then I think John's (second) set of commands needs a change to the definition of xx, as follows: xx <- exp(seq(from=log(min(x)),to=log(max(x)),length=50)) m <- lm(y ~ log(x)) yy <- predict(m, data.frame(x=xx)) points(xx, yy, col="red") HTH, Eric On Mon, Sep 25, 2017 at 11:36 PM, Fox, John <jfox at mcmaster.ca> wrote:> Dear Rich, > > Assuming that I understand what you want to do, try adding the following > to your script (which, by the way, is more complicated that it needs to be): > > xx <- 10:50 > m <- lm(y ~ x) > yy <- predict(m, data.frame(x=xx)) > lines(spline(xx, yy), col="blue") > > m <- lm(y ~ log(x)) > yy <- predict(m, data.frame(x=xx)) > points(xx, yy, col="magenta") > > The first set of commands adds a line corresponding to the points that you > plotted, which if I understand right, is *not* what you want. The second > set of commands shows how to find points along the diagonal straight line > that you plotted, given their x-values, which is what I think you want. > > If you examine the linear models fit, you'll see that they just > interpolate between the two points, albeit differently. > > I hope this helps, > John > > ----------------------------- > John Fox, Professor Emeritus > McMaster University > Hamilton, Ontario, Canada > Web: socserv.mcmaster.ca/jfox > > > > > > -----Original Message----- > > From: R-help [mailto:r-help-bounces at r-project.org] On Behalf Of Evans, > > Richard K. (GRC-H000) > > Sent: Monday, September 25, 2017 3:28 PM > > To: r-help at r-project.org > > Subject: [R] bowed linear approximations > > > > Hello, > > > > Please run the following code snippet and note the resulting plot: > > > > x <- c(10, 50) > > y <- c(0.9444483, 0.7680123) > > plot(x,y,type="b",log="x") > > for(i in 1:50){ > > xx <- exp(runif(1,log(min(x)),log(max(x)) )) yy <- approx(x,y,xout=xx, > method > > "linear") > > points(xx,yy$y) > > } > > > > notice the "log=x" plot parameter and the resulting "bow" in the linear > > approximation. > > > > This makes sense when I realized that the plot command is first making > the > > plot and then drawing straight lines between points on a log plot AFTER > the > > plot is generated and that that's why the line is straight. I get that. > > .. and it also makes sense that the bowed points are a result of the > linear > > approximations being made BEFORE plotted in a logarithmic plot. I get > that.. > > > > My goal is to make approximations that lie on the line produced on the > plot as > > shown, so I realize that what I want to do is NOT linear approximations, > but > > maybe "log" approximations? > > However, the approximation methods are only "linear" and "constant" .. > there > > isn't a "log" method to approximate with. > > > > So can anyone tell me how to fix the code such that he approximated > points > > DO lie on the line as plotted with the "log=x" plot parameter? > > Oh, and they have to be uniformly distributed along the Log=x axis.. I > think > > that's the tricky part. > > > > Any help and/or insight would be greatly appreciated. > > > > Thank you! > > -Rich > > > > > > [[alternative HTML version deleted]] > > > > ______________________________________________ > > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > > https://stat.ethz.ch/mailman/listinfo/r-help > > PLEASE do read the posting guide http://www.R-project.org/posting- > > guide.html > > and provide commented, minimal, self-contained, reproducible code. > > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/ > posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >[[alternative HTML version deleted]]
Eric, John, thank you both very much for responding.
Ok.. I suppose I need to show more of the actual data.
I have this data:
freq <- c(2, 3, 5, 10, 50, 100, 200, 300, 500, 750, 1000, 1300, 1800, 2450,
2900, 3000, 4000, 5000, 6000, 7000, 8200, 9300, 10000, 11000, 18000, 26500,
33000, 40000)
mag <- c(1.9893038, 1.5088071, 1.1851947, 0.9444483, 0.7680123, 0.7458169,
0.7069638, 0.6393066, 0.6261539, 0.6263381, 0.7053774, 0.6900626, 0.6953527,
0.7843036, 0.9056359, 0.8867276, 0.8937421, 0.9492288, 0.9629118, 1.1972268,
1.0010515, 0.9945838, 1.0564356, 0.8733333, 1.1666667, 1.5366667, 1.4666667,
1.3166667)
that must be displayed this way:
plot(freq,mag,type="b",log="x")
Essentially, I just want to show that I can reliably approximate the magnitudes
of new data for additional random frequencies uniformly lying on the frequency
axis. And so I write the following:
for(i in 1:200){
xx <- exp(runif(1,log(min(freq)),log(max(freq)) ))
yy <- approx(freq,mag,xout=xx, method = "linear")
points(xx,yy$y,col=rgb(1,0,0))
}
And I have been puzzling over why the approximated points don't lie linearly
over the original data set (especially prominent in the bow between freq=10 and
50). Once I realized (and concurred with) why this bow exists, I have been
struggling with how to make these approximations as expected.. In my original
post, I think I oversimplified it too much by implying that my application was
just 2 data points.
Are your suggestions still valid do you think?
-Rich
My apologies for the typos in the code.
Here is a corrected version you can copy/paste in R to see the issue.
freq <- c(2, 3, 5, 10, 50, 100, 200, 300, 500, 750, 1000, 1300, 1800, 2450,
2900, 3000, 4000, 5000, 6000, 7000, 8200, 9300, 10000, 11000, 18000, 26500,
33000, 40000);
mag <- c(1.9893038, 1.5088071, 1.1851947, 0.9444483, 0.7680123, 0.7458169,
0.7069638, 0.6393066, 0.6261539, 0.6263381, 0.7053774, 0.6900626, 0.6953527,
0.7843036, 0.9056359, 0.8867276, 0.8937421, 0.9492288, 0.9629118, 1.1972268,
1.0010515, 0.9945838, 1.0564356, 0.8733333, 1.1666667, 1.5366667, 1.4666667,
1.3166667);
plot(freq,mag,type="b",log="x");
for(i in 1:200){
xx <- exp(runif(1,log(min(freq)),log(max(freq)) ));
yy <- approx(freq,mag,xout=xx, method = "linear");
points(xx,yy$y,col=rgb(1,0,0));
}
For completeness, I have been puzzling over why the approximated points
don't lie linearly over the original data set (especially prominent in the
bow between freq=10 and 50). Once I realized (and concurred with) why this bow
exists, I have been struggling with how to make these approximations as
expected.. In my original post, I think I oversimplified it too much by implying
that my application was just 2 data points.
Are your suggestions still valid do you think?
-Rich
Dear Rich, I think that it's generally a bad idea to give statistical (as opposed to simply technical) advice by email without knowing the context of the research. I think that you'd do well to seek help from a statistician, and not just do what I suggest below. Interpolating the data only makes sense if there's no random component to the response (mag in your data). Otherwise, it makes more sense to get "predictions" from a statistical model that has an explicit error component for the response. In your case, a simple quadratic model in log(freq) seems to fit the data reasonably well. To see what I mean, try plot(log(freq), mag) mod <- lm(mag ~ poly(log(freq), 2)) summary(mod) points(log(freq), fitted(mod), pch=16) lines(spline(log(freq), fitted(mod))) Some basic regression diagnostics suggest that we can do better by taking the log of mag as well, producing a closer fit to the data and stabilizing the error variance: plot(log(freq), log(mag)) mod2 <- lm(log(mag) ~ poly(log(freq), 2)) summary(mod2) points(log(freq), fitted(mod2), pch=16) lines(spline(log(freq), fitted(mod2))) I have no idea whether this makes substantive sense in the context of your problem. Best, John> -----Original Message----- > From: R-help [mailto:r-help-bounces at r-project.org] On Behalf Of Evans, > Richard K. (GRC-H000) > Sent: Tuesday, September 26, 2017 10:01 AM > To: Eric Berger <ericjberger at gmail.com>; Fox, John <jfox at mcmaster.ca> > Cc: r-help at r-project.org > Subject: Re: [R] bowed linear approximations > > My apologies for the typos in the code. > Here is a corrected version you can copy/paste in R to see the issue. > > freq <- c(2, 3, 5, 10, 50, 100, 200, 300, 500, 750, 1000, 1300, 1800, 2450, 2900, > 3000, 4000, 5000, 6000, 7000, 8200, 9300, 10000, 11000, 18000, 26500, 33000, > 40000); mag <- c(1.9893038, 1.5088071, 1.1851947, 0.9444483, 0.7680123, > 0.7458169, 0.7069638, 0.6393066, 0.6261539, 0.6263381, 0.7053774, > 0.6900626, 0.6953527, 0.7843036, 0.9056359, 0.8867276, 0.8937421, > 0.9492288, 0.9629118, 1.1972268, 1.0010515, 0.9945838, 1.0564356, > 0.8733333, 1.1666667, 1.5366667, 1.4666667, 1.3166667); > plot(freq,mag,type="b",log="x"); for(i in 1:200){ xx <- > exp(runif(1,log(min(freq)),log(max(freq)) )); yy <- approx(freq,mag,xout=xx, > method = "linear"); points(xx,yy$y,col=rgb(1,0,0)); } > > For completeness, I have been puzzling over why the approximated points > don't lie linearly over the original data set (especially prominent in the bow > between freq=10 and 50). Once I realized (and concurred with) why this bow > exists, I have been struggling with how to make these approximations as > expected.. In my original post, I think I oversimplified it too much by implying > that my application was just 2 data points. > > Are your suggestions still valid do you think? > -Rich > ______________________________________________ > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting- > guide.html > and provide commented, minimal, self-contained, reproducible code.
[solved] -- As always, the solution is simple and I always feel foolish for not
having seen it right away.
Thank you, John, for showing that the frequency data can be
"converted" to log before the linear approximations are made.. once
the linear approximations are done with log data, the approximations can be
plotted into the "log=x" plot and everything is spot on :-)
Here is the updated code that does exactly what I had originally wanted:
freq <- c(2, 3, 5, 10, 50, 100, 200, 300, 500, 750, 1000, 1300, 1800, 2450,
2900, 3000, 4000, 5000, 6000, 7000, 8200, 9300, 10000, 11000, 18000, 26500,
33000, 40000);
mag <- c(1.9893038, 1.5088071, 1.1851947, 0.9444483, 0.7680123, 0.7458169,
0.7069638, 0.6393066, 0.6261539, 0.6263381, 0.7053774, 0.6900626, 0.6953527,
0.7843036, 0.9056359, 0.8867276, 0.8937421, 0.9492288, 0.9629118, 1.1972268,
1.0010515, 0.9945838, 1.0564356, 0.8733333, 1.1666667, 1.5366667, 1.4666667,
1.3166667);
flog <- log(freq);
plot(exp(flog),mag,type="b",log="x");
for(i in 1:2000){
xx <- runif(1,min(flog),max(flog));
yy <- approx(flog,mag,xout=xx, method = "linear");
points(exp(xx),yy$y,col=rgb(1,0,0));
}
Thank you!
-Rich