How can I have for loop show me each histogram it created rather then play it like a movie?

I am trying to see each histogram (with a pdf fit to it) for each value within the for loop but when i run it it is "playing" each histogram plot back to back like a short movie. I want to see each histogram and potentailly place it all in one figure if that makes sense. Here is the current code I have for the section.

"

for i=1:10;

idx_age = find(age >= (i-1)*1000 & age < i*1000);

age_tmp = age(idx_age);

iso_tmp = d18O (idx_age);

x1 = iso_tmp

% fit normal distribution to sample data

pd1 = fitdist(x1,'Normal')

mu1 = mean(pd)

sigma1 = std(pd)

% Overlay a plot of the fitted pdf.

x1_pdf = [-50:0.1:-20]

y1 = pdf(pd,x1_pdf)

% then use iso_tmp to make a pdf within for loop and save to a matrix to plot

figure (3)

histogram(x1,'Normalization','pdf')

line(x1_pdf,y1)

"

Answer 1

[Star Strider]

I am not certain what you are doing or the result you want. I don’t see an end for the for loop, so I assume everything is in the same loop.

Perhaps a subplot array ( or tiledlayout) would do what you want —

figure(3)

for i=1:10;

idx_age = find(age >= (i-1)*1000 & age < i*1000);

age_tmp = age(idx_age);

iso_tmp = d18O (idx_age);

x1 = iso_tmp

% fit normal distribution to sample data

pd1 = fitdist(x1,'Normal')

mu1 = mean(pd)

sigma1 = std(pd)

% Overlay a plot of the fitted pdf.

x1_pdf = [-50:0.1:-20]

y1 = pdf(pd,x1_pdf)

% then use iso_tmp to make a pdf within for loop and save to a matrix to plot

subplot(5,2,i)

histogram(x1,'Normalization','pdf')

line(x1_pdf,y1)

title(sprintf('Histogram %2d',i))

end

.

[Brooke Chase]

Oh WOW that was nice!

Is there potentially a way to save it as a video/clip so that it shows the change through time?

Also while you are here, is the command I am using currently creating a pdf fit to a histogram (because that is what I am trying to do - fit a pdf to a histogram)? I have been struggling with that. It looks correct but sometimes it doesn't seem correct.

[Image Analyst]

Brooke, see attached demo. It plots a surface and makes a movie of it as it changes. Adapt as needed.

[Star Strider]

@Brooke Chase —

Thank you!

See the documentation section on Animation to begin with. Then experiment. I have done some of this in the past, so I can help (to an extent).

A distribution fit to a histogram is likely never going to be exact, unless the data are plentiful, bin numbers are reasonably high (so that the bin widths are small) and the data are distributed according to the chosen distribution. (This is usually more a matter of asethetics than mathematics.) It may also be appropriate to experiment with different distributions, since for example physiological data are generally modelled more precisely (and correctly) by a logmnormal distribution than a normal distribution. The distribution also has to ‘make sense’ in the context of the data.

[Brooke Chase]

Okay, I understand what you are saying.. But when I used this loops it looks as though the pdf for each histogram remained the same? - the histogram was changing but not the pdf? any suggestions?

[Star Strider]

I do not have your data, however the automatic axis scaling may make the pdf look the same in each plot.

If you define the same axis limits for each plot, using for example:

axis([xmin xmax ymin ymax])

where the elements are able to fit the extremes of every plot, they may change between the plots. If the data in each histogram can be described by approximately the same mean and standard deviation (assuming a normal distribution, or the same parameters for other distributions), then the pdfs may not change much, if at all.

.

[Brooke Chase]

Okay so it could make sense for the pdf to remain just about the same for each histogram and look something like this?

[Star Strider]

It would, however I would expect it to line up with the histogram more exactly, unless the intent is to demonstrate that they are different.

For example —

x1 = 1.5*randn(1E+4,1)-32;
pd = fitdist(x1, 'Normal');
x1_pdf = [-50:0.1:-20];
y1 = pdf(pd,x1_pdf);

figure
histogram(x1, 'Normalization','pdf')
hold on
plot(x1_pdf, y1, '-r', 'LineWidth',2)
hold off

.

[Brooke Chase]

Yeah, that is what I thought too but that is not what I am getting. It looks like the exact same pdf for the 10 histograms I created within the loop.

[Brooke Chase]

Also, thank you for all of your help so far. I really appreciate it.

[Star Strider]

As always, my pleasure!

I don’t have your data however I believe that I see the problem:

pd1 = fitdist(x1,'Normal')

mu1 = mean(pd)

sigma1 = std(pd)

y1 = pdf(pd,x1_pdf)

See if changing those to:

pd1 = fitdist(x1,'Normal')

mu1 = mean(pd1)

sigma1 = std(pd1)

y1 = pdf(pd1,x1_pdf)

solves that problem.

I didn’t see this before because I wasn’t looking for it. My apologies for that oversight.

.

[Brooke Chase]

Ahh there it is! That worked, thanks a ton!!

[Star Strider]

As always, my pleasure!

[Brooke Chase]

One more question.. so sorry. How would you suggest I plot all of the histograms together on one plot? - create a plot together with a color bar. Trying to show a change through time but in one figure (for a proposal)

[Star Strider]

No worries! (I was off doing other things for a while.)

Puitting the pdf’s together without the histograms would suggest a waterfall or ribbon plot.

Plotting histograms and pdf’s together in one axes would likely not be possible without hopelessly confusing all of them. For that reason,.I would put the 10 plots together in one figure either using subplot (similar to what I posted earlier) or tiledlayout. Considering that they all share the same independent variable, stackedplot might also be an option. With all of these, I would use the same axis call (as I outlined earlier) to make differences between them readily apparent, instead of using automatic axis scaling that would obscure the differences.

Use saveas or similar functions to save the figure as an image.

Answer 2

[Paul]

What happens if you put the command

hold on

after the figure(3) line?

[Brooke Chase]

Okay that helped a little because it color coded each plot in the loop rather than the end result being the last value in the loop. I guess I want a better way to visualize the change because right now I am just seeing a mush of all the plots on top of one another? So maybe a different plot for each value or something?