Merging two arrays containing decimal points into a table

I am currently working on data stored in netCDF format.

Each variable was read into a variable (n x 1 array) using the ncread() function

Ex. var = ncread(FilePath, 'var')

These variables contain decimal points (Ex. -1.0874020).

Using these individual variables I have read in, I would like to construct a table.

Ex.

data = [var1 var2 var3];

colNames = {'var1', 'var2', 'var3'};

table= array2table(data, 'VariableNames', colNames);

However, when I join the variables together, (data = [var1 var2 var3];) for some reason the variables stored in 'data' turns into integers (Ex. -1).

How can I merge arrays containing decimal points into a table?

Thanks,

[dpb]

Show us all the code in context and the result of

whos var* data
format short;format compact
head(table)

Unless the data are not numeric, any apparent loss of precision will simply be one of how the data are shown on screen; MATLAB will not have rounded nor truncated a double in the process of catenating variables; something else is confusing the picture here...but we don't have enough enough to know just what, not having the actual data nor precise code in entirety.

[jessupj]

i agree with @dpb here. you might also check that the variables in the netcdf aren't single precision.

table= array2table(double(data), 'VariableNames', colNames);

might be worth a quick try

[George Sato]

Hey,

Thank you for the quick response.

I am analyzing data collected from a tag attached to an animal.

The data logger collected the three-axis specific acceleration (accel), and the three-axis magnetic field (mag).

Sorry about that, I will show some of my code here.

% File path for the netCDF files (accel, mag data)
filePath = '/Users/~/filename.nc'

% Read the netCDF file into MATLAB
ncdisp(filePath)

% Read one of the variables from the netCDF file
order = ncread(filePath, 'order'); % This is the frequency of data collection
accel_y = ncread(filePath, 'accel_y'); % Accelerometer data from the y-axis
mag_y = ncread(filePath, 'mag_y'); % Magnetometer data from the y-axis

% Combine the arrays together
data = [order accel_y mag_y];
colNames = {'order', 'accel_y', 'mag_y'};

% Convert data into a table
accel_data = array2table(data, 'VariableNames', colNames);

For the answer to your codes, here it is:

whos var* data
% Name           Size              Bytes  Class    Attributes

%  data      127516x3             1530192  int32 

format short;format compact
head(accel_data)
% ans =
%  8×3 table
%    order    accel_y    mag_y
%    _____    _______    _____
%      1        -1          0 
%      2        -1          0 
%      3        -1        -71 
%      4        -1        -71 
%      5        -1        -72 
%      6        -1        -72 
%      7        -1        -72 
%      8        -1        -72 

For the array that I read the netCDF data in, the results were the following:

whos var* accel_y
%   Name              Size             Bytes  Class     Attributes

%  accel_y      127516x1             510064  single   

The data is too large for me to show it to you here.

It seems that when I combine the two columns, the Class of data becomes int32. Is this fixable?

Thanks,

Answer 1

[dpb]

Aha! As we suspected, the problem is in what was NOT shown -- the order variable is apparently an int32; MATLAB will silently coerce variables of different numeric classes to the lowest common denominator -- observe

>>  [int32(rand()*199) single(rand)*100 double(rand)*123456]
ans =
  1×3 int32 row vector
      102       75   104447
>> 

That's what you've done above. Convert the individual variables to the table; there's no need to combine them into an array and then just split it back up again --

...
colNames = {'order', 'accel_y', 'mag_y'};
% Convert data into a table
accel_data=table(order,accel_y,mag_y, 'VariableNames', colNames);
...

[George Sato]

dpb Thank you so much!

'the order variable is apparently an int32; MATLAB will silently coerce variables of different numeric classes to the lowest common denominator '

This makes so much more sense!

As you have mentioned, I think I will directly turn it into a table!

Thanks!

[George Sato]

Hey dbp,

I tried the suggested code:

colNames = {'order', 'accel_y', 'mag_y'};
% Convert data into a table
accel_data=table(order,accel_y,mag_y, 'VariableNames', colNames);

However, it gives off the following error:

Error using array2table:

Invalid parameter name. Parameter name must be a nonempty string or character

vector.

accel_data = array2table(order,accel_y,mag_y, 'VariableNames', colNames);

Do you know any other way to construct a table?

Thanks,

[dpb]

The syntax is fine; it's in something else instead you've done that we don't have code/data for in complete context to see the whole ball o' wax...

But, if I create variables of the given types, it creates the table as shown just fine--

order=int32(1:10).';
accel_y=single(randn(size(order)));
mag_y=single(randn(size(order)));
colNames = {'order', 'accel_y', 'mag_y'};

tA=table(order,accel_y,mag_y,'VariableNames',colNames)
tA =
  10×3 table
    order     accel_y        mag_y   
    _____    _________    ___________
      1       -0.30994       -0.59436
      2       -0.58999         1.2665
      3       -0.71516       -0.22809
      4         1.0615    -0.00097881
      5       -0.24335         1.7408
      6       -0.18695         1.2983
      7       -0.54987       -0.12946
      8        -1.8558         1.6106
      9        0.65487        -1.2138
     10      -0.028611         1.8246
>> 

As can see, the syntax is fine if the data are and there are no typos, etc., ...

OH! You've still called array2table instead somewhere is what caused the error -- get rid of that entirely; that's simply wrong.

accel_data = array2table(order,accel_y,mag_y, 'VariableNames', colNames);

it's table with the different variables;

Read <table> and <array2table> documentation to see the difference in functionality and syntax...

[George Sato]

THANK YOU SO MUCH!

It worked!

So I should be using 'table' instead of 'array2table'

Thank you so much!