fitline not displaying correct x values datetime

I an attempting to plot a linear OLS regression line from a timetable variable that I have, however somewhere along the way the x values are getting distorted so when I plot the graph they arent at the correct date/datenum (extremely far off).

Here is what I have so far:

dates = datenum(Timetable.Time);
myfitD = polyfit(dates, Timetable.Var1, 1);
fitline = polyval(myfitD, dates);
plot(fitline);

I need to be able to add this line to an existing plot where the X axis is in datetime format.

"Timetable variable" has dates ranging from 01-Jan-2015 to 01-Jan-2099, but based on the graph its currently producing it looks to me as if the fitline is showing the correct y values at an x range of [15 90] which I dont know how to convert to be able to add to this other graph.

I've used this same method before with a similar dataset and it worked just fine, so I'm not sure what the issue is.

For additional context, the graph I am attempting to add it to is plotted as follows:

Years = {'01-jan-2030'; '01-jan-2050'; '01-jan-2070'; '01-jan-2090'};
Xaxis = datetime(Years);
scatter(Xaxis, Yvar);
x1 = datetime('01-Jan-2020');
x2 = datetime('01-Jan-2099');
xlim ([x1, x2]);

Answer 1

[Star Strider]

You probably need to do centreing and scaling with your data, then use datetick to display the dates correctrly —

dates = datenum(Timetable.Time);
[myfitD,S,mu] = polyfit(dates, Timetable.Var1, 1);
fitline = polyval(myfitD, dates, S, mu);

figure
plot(dates, fitline)
datetick('x', 'keepticks', 'keeplimits')

See the dateFormat documentation section for those details.

.

[Peter Perkins]

datenums and datetick are not the best thing to use, but there is a version of the same idea that would use datetime: get dates as something like

dates = Timetable.Time - mean(Timetable.Time);

and then you should just be able to add Timetable.Time,fitline to the scatter plot that used a datetime axis.