how to increase the weight of a certain point when using smoothing splines?

how to increase the weight of a certain point when using smoothing splines?

the certain point is the ionosonde and I want to pass the line fittingcurve through this point

load('data.mat');

rows=size(RO_ISR,1);

for i=1:rows

RO=RO_ISR{i,1};

IONOSODE=RO_ISR{i,2};

x1 = RO.EDP8 * 10^6;

y1 = RO.EDP3;

columns=size(RO_ISR,2);

for j=3:columns

ISR=RO_ISR{i,j};

x2 = ISR.NE8;

y2 = ISR.GDALT;

TT1 = table2timetable(ISR);

TT3 = synchronize(IONOSODE, TT1);

F = fillmissing(TT3,'previous');

ISR_date =ISR.Date;

F_date = F.Date;

[logical_Index ,index] = ismember(ISR_date(1),F_date);

x3=IONOSODE{index,10};

y3=IONOSODE{index,7};

y4 = [y2;y3];% add row y3 to column y2

final_altitude=sort(y4); % sort data in asending form

[logical_Index ,index]= ismember(y3,final_altitude);% find index of y3 in final_altitude

final_ElectronDensity= [x2(1:index-1)', x3, x2(index:end)'];

final_ElectronDensity=final_ElectronDensity';

fittingcurve = csaps(final_altitude,final_ElectronDensity,0.0003,final_altitude);

figure;

hold on;

plot(x1, y1);

plot(x2, y2);

plot (x3, y3,'-s','MarkerSize',9, ...

'MarkerEdgeColor','blue', ...

'MarkerFaceColor',[1 .4 .4])

plot(fittingcurve,final_altitude,'g');

xlabel('Electron Density');

ylabel('Altitude');

title('Plotting');

legend('RO', 'ISR','ionosonde', 'fitting curve');

hold off;

end

end

[t]

any idea to solve this question?

[John D'Errico]

A bit impatient, are you? You might recognize that you asked a question at a time that would have been in the middle of the night for all of the US, where many of the people live who might be able or willing to answer your question.

Answer 1

[John D'Errico]

Just read the help for csaps. Here is a link for the doc: csaps

In there, I find this line:

pp = csaps(x,y,p,[],w) also specifies the weights w in the error measure

HOWEVER, just increasing the weight on a point will not force the curve to pass exactly through that point. It will reduce the degree the curve misses that point. But not exactly force the fit. And if you too massively increase one weight, this may also introduce numerical problems in the fit.

You might decide to use other tools, like my SLM toolbox, which can do smoothing, as well as force a fit through any given point or set of points. You can find SLM for free download from the File Exchange.

[t]

I did not understand the doc in the link, but I tried this

pp= csaps(final_altitude,final_ElectronDensity,0.2,[ones(1,10),repmat(5,1,10), 0]);