Time Series analysis problems with date

Hi all-and a happy new year (lets hope its better than the last one......)

My query concerns time series.

I have a dataset which consists of 7 seperate excel files, first column is date and second column is a scalar number.

The date is a period of 8 months, jan-aug for 10 years (repeated in each excel file).

I have plotted the first file and the first 8 months plot fine for the ten year period, however, the sep-dec period is also plotted which shows no data, therefore there is a straight line until the next year starts on the graph.

I have used the code

= datevec(cam{:,1}); and also tried

= datenum (cam{:,1});

both do not work!!

So my question is how do I plot the data for just the dates shown which Jan 1st-Aug 31st 2010-2020?

Im fairly new to all this, I have tried using the mathworks help section and tried several other pieces of code, but I get the dreaded red writing, I apologise if this is a billy basic question!

many thanks in advance

Mike

[Eric Sofen]

It will be helpful to see some sample data and more of the code you've tried to write for context. I'm not sure what cam{:,1} refers to.

To get things started, I'd encourage you to use the datetime type rather than the older datenum/datevec.

[mike]

Hey Eric,

sorry for the delay in response!

Column 1 are the dates and column 2 are the values I am interested in.

% this is how I load the data
infmt = 'dd-MM-yyyy';
data_all = datetime(testData{:,1}, 'InputFormat',infmt); 
gas = testData{:,2};

I now tried to replace the values from Sep-Dec with NaN, but this tricks Matlab into thinking that these values are equal to dates where there are no measurements (eg system failure). So I can't do interpolation etc.

I then tried to concatenate (see below), but this somehow looses all the date information.

n2010 = find(data_all>='2010-01-01' & data_all<='2010-08-31');
n2011 = find(data_all>='2011-01-01' & data_all<='2011-08-31');
C1=cat(1,n2010,n2011);

Is there a way to only select/plot Jan-Aug over a set period of years?

Thank you!

[Adam Danz]

Is the goal to remove the straight line from the data while maintaining the sept-dec dates along the x axis? Or is the goal to remove the gap completly and jump from Aug to Jan along the x axis?

[mike]

Hi Adam,

Ideally to remove the lines and the gaps from the x axis so that I can them smooth and detrend.

Cheers

Answer 1

[Biral Pradhan]

I understand, you want to remove the straight line as well as x-axis labels corresponding to those 4 months (Sep-Dec) for each year. I assume, you plan on interpolating the data for these 4 months later. While plotting monthly data using datetime values from different tables (as in your case), the axis spacing is determined uniformly, so you get labels separated by 1 calendar month.

I do not think it is possible to abruptly change the spacing from 1 calender month to 4 abruptly. A workaround would be to use a timetable and resample your data.