How to convert continuous variable into discrete variables?

I am looking for the help !!

I have 5000 rows of data points for three variables (independent variables-X1,x2,x3 and dependent variables Y). The x1 and x2 are discreate variables in fix steps. However, the X3 variables is countinous ranges from 0-0.6. I want to transfrom this varibles in steps of 0.05 upto 0.6 total in 12 steps and corroespoing value of Y variables should be mean of all values correspond between 0- 0.05 and so.

Well, i tried with find(x3<0.05) and got the index of all values which is less than 0.05. but struggling to compute mean(y) for the indices, mostly they are continous.

Plz help me.

Answer 1

[Chetan Badgujar]

HI Alan,

Thank you for you reply,

I tried but we are losing some information there..I attached the plot comparsion before and after converting the data, discreate data should follow the countinous data form.

load('DATA.mat');

X3(1) = []; % the first values for X3 and Y are NaNs, so they are removed here

Y(1) = [];

x3 = (0:0.05:2)';

Ydiscrete = zeros(length(x3),1);

for i = 2:length(x3)

ix = find(x3(i-1)<=X3<x3(i));

Ydiscrete(i) = mean(Y(ix));

end

plot(x3,Ydiscrete,'o')

hold on

scatter(X3,Y)

[Alan Stevens]

Hmm. What about the following:

load('DATA.mat');
X3(1) = [];  % the first values for X3 and Y are NaNs, so they are removed here
Y(1) = [];
x3 = (0:0.05:2)';
delta = length(Y)/41;
Ydiscrete = zeros(length(x3),1);
for i = 2:length(x3)
    ix = (i-1)*delta:i*delta;
    Ydiscrete(i) = mean(Y(ix));
end
subplot(2,1,1)
plot(X3,Y,'.',x3,Ydiscrete,'o')


Y = sort(Y);
X = 1:length(Y);
delta = length(Y)/41;
x = 1:delta:length(Y);
for i = 2:length(x)
       ix = (i-1)*delta:i*delta;
    Ydiscrete(i) = mean(Y(ix));
end
subplot(2,1,2)
plot(X,Y,'.',x,Ydiscrete,'o')

where the top plot uses X3 and Y, but the bottom plot basically just sorts Y.

:

[Chetan Badgujar]

HI Alan,

Well, I manuuly tried to compute this and explain very well in attached excel sheet Please go through it. The manual calculation sumarize it better way.

so WHat i did,

Step 1) find 0.05 in x and do average y( all values between 0-0.05)

Step 1) find 0.1 in x and do average y( all values between 0.05-0.1)

--- and so on until last value of x

It gives a great result you can check the excel sheet graph.

It would be great help even if You can convert all above steps into code.

Thank you once again.

Answer 2

[Alan Stevens]

Does the following do what you want?

X3 = 0:0.05:0.6;
Ydiscrete = zeros(length(X3));
for i = 2:length(X3)
    ix = find(X3(i-1)<=x3<X3(i));
    Ydiscrete(i) = mean(Y(ix));
end

Untested, as I don't have the data!

[Chetan Badgujar]

Hey, Alan Thanks for you reply.

I tired the given code, but it does not work. Well I have attached the data. Could you please test on that?

Thank you..

Answer 3

[Alan Stevens]

How about this

load('DATA.mat');
X3(1) = [];  % the first values for X3 and Y are NaNs, so they are removed here
Y(1) = [];
x3 = (0:0.05:0.6)';
Ydiscrete = zeros(length(x3),1);
for i = 2:length(x3)
    ix = find(x3(i-1)<=X3<x3(i));
    Ydiscrete(i) = mean(Y(ix));
end
plot(x3,Ydiscrete,'o')

Answer 4

[Cris LaPierre]

No need to ask the same question multiple times.

[Chetan Badgujar]

Hi @Cris,

I am using Matlab R2018A and getting following error

Error using groupsummary (line 103)

First argument must be a table or a timetable.

Error in Untitled (line 8)

Ymean = groupsummary(Y,X3,rg,@mean,"IncludeMissingGroups",false)

Yes, you are right I just want to summaries data, convert all rows of 5000 in to whatever discreat steps its coming.

Thanks.

[Chetan Badgujar]

@Cris,

It works well, but when I compare Y with newY.. Code

plot(Y)

hold on

plot(newY)

The NewY doesnot represent the Y.

You can check by plotting both. I feel like there is some data leakage.

Thanks

[Chetan Badgujar]

@ Cris,

Thank you very much.. This is what I was looking for.. Its perfect with manual calculations.

Thank you once again.!!!