Dear Amin Sadeqi,

If your raw data are in csv or xlsx file, you can input them into Matlab.

Then, Matlab will resave your data as a matrix or cell.

At this moment, I think you can use many useful functions like

find() to return the row number of missing data (null cell), or logical judgement command: a=NaN

At last, you can easily count the number of missing data.

Hope this can help you.

Best Regards.

Yours, sincerely.

I suggest to review this paper:

Article Comparison of different Methods for Univariate Time Series I...

I hope this helps

Best

Ramon

A sufficient sound, quick and easy method to estimate missing data could be LOESS (locally weighted regression smoother)

As it seems you are familiar with Matlab, I suggest to read this page " Data Driven Fitting with MatLab" https://www.mathworks.com/matlabcentral/fileexchange/31562-data-driven-fitting-with-matlab

Best

Ramon

Dear Kai Liu

My main problem is not missing data counts. My main problem is estimating missing data.

Thank you for your attention

Thank you Mohammad. I do not have any information about multiple imputation by chained equations .

Using the EM algorithm is a good way to estimate missing data. Or maybe inputting the mean/average but understand that is an underestimate of variance.

Dear Amin

Sorry for misunderstanding your question. As Mohammad Yusuf pointed out, you can use multiple imputation approach for estimating missing data. Maybe, in my opinion, you can input the mean/average variance like Vivian Madu's answer. Anyway, if you have known where the missing data are and what the estimating rule is, you can use loop command for completing the calculation quickly through Matlab or other software. After using find() function for identifying the row number of missing data, then I think it is possible to perform the estimating process by using loop command.

By estimate do you mean predict? Are you looking to fill the missing gaps or determine the volume of data that is missing?

Dear Cary

I'm looking to fill the missing gaps

Dear Amin,

A quick and simple approach for filling the missing data points is to apply a moving average including 3 previous and 3 next data points.

This is a very practical methods, but it has limitations when there are multiple continuous missing data points.

I hope this helps.

Dear Amin,

I wish you also see into the approach proposed by Sir Ramon Martinez.

I have one used that and the paper was accepted.

Following!

It is not a recommended to take a quick and dirty path to fill the gaps. Although imputation seems a logical step to fill the data, a deeper understanding of the data, distribution of the sample and population must be effectively considered before attempting the gap filling procedure. In other words, missing data parts could be your local minima or maxima which you would misrepresent if you use a simple imputation approach. Filling missing data must be done case by case, with good understanding of the data. Compare the data with surrogate or similar studies to collect evidence. So shrug the urge to fill the gaps with quick method and focus on the objectives, distribution of data before choosing your path.

I suggest you to use hydrolab (https://hydrologie.org/MOD/Hydrolab/hydrolab.htm)

Thank you Zerouali Bilel , but it is in French. Can you introduce to me an English version?

Dear @Amin sadeqi, I think that Hydrolab available only in french version

check this discussion to find another solution please (https://www.researchgate.net/post/How_might_I_fill_missing_data_of_river_flow_and_temperature_data_hydrological_data)

Dear Amir

you can use a reference station and using a regression analysis fill the gaps. Also using the EM algorithm may help you but it is somehow elaborating. if you have a dense sattion network, you can use GIS. By spatial analyst it is possible to fill the missing data.

Bests,

Interpolation will help to know missing Data

Hi,

I don't know if this method work for you or not? I tell you anyway. Personally in some cases I used curve fitting methods. There are discontinued points in every data sheet. By various curve fitting methods available throughout the internet you can fix this issue.