how to label an array having a range of values with only two texts

I have a range of values in an array such as (1:89) and I want to label them as either oranges or apples depending on a threshold assigned in their PDF??

[Chad Greene]

This question is not very clear. What exactly do you want to do? What do you mean by label?

[BR]

I want to create a data set like Fisher iris dataset. But the division of colors should be on the basis of a threshold assigned in the PDF for the datset (as an example 1000 rand values between 1-50).

Like the values above a certain threshold should be labelled as RED and the values below the threshold as ORANGE.

Answer 1

[Walter Roberson]

YourArray = randi(89, 1, 1000);

magic_threshold = 0.01381972;
[uvals, ~, uidx] = unique(YourArray);
portions = accumarray(uidx, 1) ./ length(YourArray);
is_orange = portions < magic_threshold;
cmap = [255 165 0; 141 182 0] ./ 255;   %orange; apple green
item_type = 2 - is_orange(uidx);  %true for orange maps to index 1, false maps to index 2;
item_color = cmap(item_type, :);
dot_size = 24;
x = 1 : length(YourArray);
scatter(x, YourArray, dot_size, item_color);
orange_mask = item_type == 1;
text(x(orange_mask), YourArray(orange_mask), 'orange');
text(x(~orange_mask), YourArray(~orange_mask), 'apple');

The output is nearly unreadable because of all of the text objects, which would tend to suggest that labeling might not be the best of ideas.

[BR]

Thanks Walter for the code. I am looking for a divide in the PDF as well.

Will accumarray help??

[Walter Roberson]

This is divided in the PDF. Their probabilities are their frequencies (number of occurrences) relative to the total count. The items are separated by 1, so their density does not need to be explicitly taken into account.

[BR]

Thanks!! Is there a way that I could make it just like Fisher iris dataset. Like Fisher iris data has a set of values with species as text assigned to them..

[Walter Roberson]

Fisher Iris relies upon multiple continuous coordinates, not upon scalar arbitrary values such as 1:89.

Perhaps you want to think about measurements such as standard deviations relative to the mean. Or perhaps you should consider some sort of multimodal detection; see https://stats.stackexchange.com/questions/5960/how-to-identify-a-bimodal-distribution

[BR]

Yeah you're right. That was just as example of asking. But how to assign texts like Fisher Iris dataset to different values in Matlab??

I mean if I get the continous seperation of a data set based on the distribution, how will I assign different texts like in FisherIris dataset.

Sorry for so many questions, but thanks for all.

[Walter Roberson]

You generate some measure of the separation, somehow, and then you test it against whatever threshold, creating a mask similar to is_orange that tells you which group each individual item belongs to. Then you follow along with the rest of the logic I show above to display the items and to text() the appropriate label into place.