fit the y=mx+c in time series

Hi. I have 420x1 data of time series.

How can I fit y=mx+c line in the plotting?

Answer 1

[Star Strider]

That could be a challenge.

One option is to use mldivide,\:

dt = datetime('now')+days(1:10);
T1 = table(dt(:), rand(10,1)+(0.1:0.1:1).');

B = [day(T1.Var1) ones(size(T1.Var1))] \ T1.Var2;
LinFit = [day(T1.Var1) ones(size(T1.Var1))] * B;

figure
plot(T1.Var1, T1.Var2, 'p')
hold on
plot(T1.Var1, LinFit, '-r')
hold off
grid

Depending on the time series, the results could appear a bit unusual.

.

[Nurul Ain Basirah Zakaria]

the way i plot is:

my_data: 420x1

and then i,

t1 = datetime(1982,01,01);

t2 = datetime(2016,12,31);

date=t1:calmonths(1):t2;

plot(date,my_data)

can I fit the y=mx+c and determine the threshold?

[Star Strider]

It would be necessary to use serial data numbers to fit all those data to a linear fit, so something like:

tv = datenum(date);
B = [tv(:) ones(size(tv(:)))] \ my_data;
LinFit = [tv(:) ones(size(tv(:)))] * B;

Then it would be necessary to plot it (after hold on) as:

plot(tv, LinFit)

I have no idea what ‘the threshold’ is, so I would need to know that.

[Nurul Ain Basirah Zakaria]

ya im sorry. its not the threshold ;) thank you very much!

[Star Strider]

My pleasure!

Did I provide a meaningful solution?

If not, what do you want to do?