Issue regarding a data plot

Hi,

I have a data as attached herewith.

The blue plot represents the gradient while the orange plot represents the percentage difference between the two datapoints.

But as you can see that at distance 0.2 the orange plot shoots up although the blue plot is sufficiently smooth.

I don't see any noise or outliers present at that point.

Can you advice me a way how to correct the percentage profile in this case?

[Divyajyoti Nayak]

Hi @Rahul, could you share the data or code you are using to plot this graph? That would help in debugging this.

[Rahul]

Hi Divyajyoti,

Sure,

Plz find it as below

intensity_gradient = data.variable.gradintensity(end,:);
Normalized_intensity_gradient = data.variable.gradintensity(end,:)./abs(min(data.variable.gradintensity(end,:)));
grid on;
yyaxis left;
plot(data.variable.x,Normalized_intensity_gradient(end,:),'.', 'MarkerSize',10); %222
xlabel('Normalized radius');
ylabel('Normalized intensity gradient');
hold on;

for i = 1:length(data.variable.x)-1
    percentage_difference_gI(i) = abs(Normalized_intensity_gradient(end,i+1)-Normalized_intensity_gradient(end,i))*200/abs(Normalized_intensity_gradient(end,i+1)+Normalized_intensity_gradient(end,i));
    
end

yyaxis right;
plot(data.variable.x(1,1:end-1),percentage_difference_gI(1,:),'.-', 'MarkerSize',10); 
ylabel('% different point');
ylim([0,300]);
yticks(0:20:300);
line([0 1],[100 100], 'LineStyle','--','Color','black'); %line([x1 x2],[y1 y2])
txt1 = text(0.025, 110, 'H-mode criteria (% different point ≥ 100 %)', 'FontSize', 11);
[Value_percentage, index_percentage] = max(percentage_difference_gI);
txt2 = text(0.025, 90, strcat("Maximum of % different point = ", num2str(Value_percentage), ' %'), 'FontSize', 11);

Answer 1

[Divyajyoti Nayak]

Hi Rahul, I think the reason your graph shoots up so suddenly is due to the way you have defined the ‘percentage_difference_gI’ vector.

percentage_difference_gI(i) = abs(Normalized_intensity_gradient(end,i+1)-Normalized_intensity_gradient(end,i))*200/abs(Normalized_intensity_gradient(end,i+1)+Normalized_intensty_gradient(end,i));

At distance 0.2, the gradient intensity is very close to 0. It is possible that the gradient intensity is crossing over from positive to negative. If this is the case then the numerator of your percentage will become larger than the denominator. For example,

Normalized_intensity_gradient(end,i+1) = 0.015;

Normalized_intensity_gradient(end,i) = -0.01;

percentage_difference_gI(i) = abs(0.015 – (-0.01))*200/ abs(0.015 + (-0.01));

percentage_difference_gI(i) = 0.025*200/0.005 = 1000

Here’s some code to show this visually with some dummy data:

data = -0.2:0.01:0.2;
percentage = abs(data(2:end)-data(1:end-1))./abs(data(2:end)+data(1:end-1));
percentage = percentage * 200;
plot(data(2:end),percentage);

Answer 2

[Rahul]

Hi Divyajyoti,

I tried to do the correction as follows

percentage_difference_gI(i) = abs(Normalized_intensity_gradient(end,i+1)-Normalized_intensity_gradient(end,i))*200/abs(Normalized_intensity_gradient(end,i+1)+Normalized_intensity_gradient(end,i));
      if Normalized_intensity_gradient(end,i)<0
        Normalized_intensity_gradient(end,i)=abs(Normalized_intensity_gradient(end,i));
      end 

But, this doesn't work. Can you advice please any other logic which can help.

[Divyajyoti Nayak]

Hey @Rahul, I'm not sure what your aim is exactly. Your expression for percentage difference is finding the percentage with respect to the average of 2 data points not the percentage increase from one data point to another. According to me, a simple percentage expression would be something like this:

percentage_difference_gI = (Normalized_intensity_gradient(end,2:end)-Normalized_intensity_gradient(end,1:end-1))./Normalized_intensity_gradient(end,1:end-1);
percentage_difference_gI = abs(percentage_difference_gI)*100;