Scatter plot and curve fitting?

Hello every one,

I have two set of vectors

X=[2,3,2,2,1,4,1,1,1,1,2,4,3,4,3,3,2,2,3,2,3,1];
Y=[9.14,38.66,0.72,0.79,1.96,0.68,7.58,0.83,2.37,3.44,2.17,1.6,21.98,5.19,0.71,25.35,17.65,7.78,14.38,4.77,2.5,29.26];

To find the correlation between X and Y, we have used Spearman rank correlation coefficient

[RHO]=corr(X.',Y.','type','spearman');

For fitting line to the data points, we have used polynomial of degree 1,

 p=polyfit(X,Y,1);%This will generate the coefficients of polynomial of degree(1)
 y=polyval(p,X);% This will give the fitted  values for the desired values.
 scatter(X,Y,'*');% This will create a scatter plot
 hold on;
 plot(X,y,'k-');% This will plot regression line.

The Spearman rank correlation coefficient has negative value, but the regression line seems to represent positive correlation between X and Y. Any help regarding this issue will be highly appreciated.

Thanks

[KSSV]

what is the significance of coor between x and y? They are different parameters right? X is your independent variable and y is your dependent variable? The following makes a sense:

RHO=corr(Y.',y','type','spearman');

[abdur rauf]

@KSSV ,

X and Y are different parameters. X is independent variable and y is your dependent variable. The correlation coefficient is non significant with P value of 0.96. I want to find the correlation between X and Y, not between Y and the fitted values. Thats why

RHO=corr(Y.',y','type','spearman');

doen't make sence to me.

My main problem is that the correlation is negative but the regression seems to show positve correlation.

Thanks

[Rik]

They are both fairly close to 0. It wouldn't surprise me if there is a lot of overlap between the two confidence intervals.

[Rik]

Comment mistakenly posted as answer by Wen Wen:

I had similar problem. I guess it's because the spearman corrlation which transformed the orignial data into ranks while the polyfit function finds the best linear fit based the original data. If this assumption is correct, my question would be how could we plot the fitting line for spearman correlation in matlab.