How to remove a year from a dataset without having the value 0

I have a 30-year long data set where two of the years (2009 and 2010) have almost no data. How can I remove these two years without having it for example affect the linear trend?

Help is grealty appreciated.

-Kristine

[dpb]

What's actually plotted in the figure--average or somesuch over each year as a single point? Can't really tell since there are no markers.

I don't see much trend of any sort; linear or not.

If you don't want to consider a couple of years, simply remove those data from the set but keep the actual year--you'll have a hole but that's just the way time is; "it keeps everything from happening at once".

ADDENDUM

That is, remove both the time and the response variables for those years not considered; shorten the overall series by that number of observations, don't just fixup the observations and leave the time values.

[Kristine]

The figure is of daily wind direction from NE (22.5-67.5 degrees). Okay, I'll try that. Thank you!

Answer 1

[Andrew P]

There are quite a few methods for regression with missing data:

http://www.statisticalhorizons.com/wp-content/uploads/MissingDataByML.pdf

You could use something like 'ecmlsrmle' which performs least square regression on datasets with missing values, where the missing values have been replaced with NaNs:

http://www.mathworks.com/help/finance/ecmlsrmle.html

[Kristine]

Just one more question. I really do not understand what I should input for Data and Design, could you explains this for me?

-Kristine