Help using the arrayfun() function to apply strsplit() to all entries in a string array

I'm trying to wrap my head around how the arrayfun() function works and would greatly appreciate some help with a specific example:

I have a string array of weather data.

weather_strings = 

  10×1 string array

    "UTC,2140991,49.0"
    "UTC,2140992,49.1"
    "UTC,2140993,49.1"
    ...

I need to extract the values after the second comma (temperatures) as a 1x10 matrix of doubles, [49.0, 49.1, 49.1, ...].

I've figured out a clunky way to do this for a single entry (please let me know if there's a better way).

weather_string = weather_strings(1) % extract only the first entry
weather_string_split = strsplit(weather_string, ',') % apply strsplit() to split on commas
weather_string_split_trim = weather_string_split(:,3) % extract only 3rd column
weather_num_trim = str2num(weather_string_split_trim) % convert from string to double

But I can't seem to figure out how to use arrayfun() to apply that to every entry. I've tried:

weather_strings_split = arrayfun(strsplit(weather_strings,','), weather_strings) % apply stringsplit to split on commas, for all elements?

which gives the error message:

Error using strsplit (line 80)
First input must be either a character vector or a string scalar.

Error in test_window (line 17)
weather_strings_split = arrayfun(strsplit(weather_strings,','), weather_strings)

I'm probably missing something painfully obvious. What is it? I'm still somewhat of a beginner at coding, so I welcome you to explain it to me like I'm 5 years old.

Alternatively, if there's a clever way to extract these numbers directly from this data table (which came directly from a webread() function), I'd love to hear it. Var3 is a cell array.

weather_data_table =

  10×3 table

       Var1         Var2             Var3       
    __________    ________    __________________

    2018-11-26    17:41:25    'UTC,2140991,49.0'
    2018-11-26    17:42:27    'UTC,2140992,49.1'
    2018-11-26    17:43:28    'UTC,2140993,49.1'
...

Again, the goal is to get just the last numbers after the second comma of Var3 into a 1D matrix.

Thanks in advance!

Answer 1

[Star Strider]

Try this:

for k1 = 1:size(weather_strings,1)
    Col3(k1,:) = str2double(regexp(weather_strings{k1}, '\d*\.\d*', 'match'));
end

Col3 =
   49.0000
   49.1000
   49.1000

The loop is necessary because regexp is not vectorised. It can only handle one srting at a time.

[Stephen23]

"The loop is necessary because regexp is not vectorised. It can only handle one srting at a time."

According to the regexp documentation the first input may be "specified as a character vector, a cell array of character vectors, or a string array. Each character vector in a cell array, or each string in a string array, can be of any length and contain any characters." I often use regexp with cell arrays containing multiple different char vectors, I don't see any reason why it should not work with string arrays, just as its documentation states it does.

>> C = {'UTC,2140991,49.0','UTC,2140992,49.1','UTC,2140993,49.1'}
>> regexp(C,'\d*\.\d*', 'match','once')
ans = 
  '49.0'
  '49.1'
  '49.1'

[Star Strider]

The regexp call threw an error with the column vector when I tried it. It works with a row vector, and OP may not want to re-format the column vector to a row vector. Also, while arrayfun definitely has its uses, when I’ve used it for problems like this, it’s been significantly slower than a simple loop, which surprised me. Thus, the loop.

[Colin Wilson]

Awesome, thank you so much Star Strider and Stephen Cobeldick! This works brilliantly for my temperature data.

Is there a way to write a similar regexp function that would isolate the number from the end of the line, regardless of whether or not it contains a decimal point? (Which is why I originally tried to use commas as delimiters.)

I also need the same function to clean up my Humidity data, which has whole integer values.

For example:

weather_strings = 
  10×1 string array
    "UTC,2140991,59"
    "UTC,2140992,61"
    "UTC,2140993,60"
    ...

If the user selects Humidity data instead of Temperature data right now, I get the following error message:

Unable to perform assignment because the indices on the left side are not compatible with the size of the right
side.

Error in clean_data (line 14)
    clean_weather_strings(:,k) = regexp(weather_strings{k}, '\d*\.\d*', 'match');

Error in Lab7 (line 23)
clean_weather_doubles = clean_data(weather_data_table) % give input to clean_data function, save output

I assume this is because our '\d*\.\d*' expression looks for digits separated by a period. I'm just not familiar enough with the syntax of the regexp() function to know how to set it up differently.

Thanks again!

[Star Strider]

My pleasure!

‘I assume this is because our '\d*\.\d*' expression looks for digits separated by a period.’

Correct.

‘I'm just not familiar enough with the syntax of the regexp() function to know how to set it up differently.’

The regexp funciton can act ‘logically’, so giving it a choice as to use the '\d*\.\d*' or '\d*', it will choose the correct pattern, with ‘|’ designationg a logical ‘or’.

To accomodate both, the regexp call changes to:

for k1 = 1:size(weather_strings,1)
    Col3(k1,:) = str2double(regexp(weather_strings{k1}, '\d*\.\d*|\d*', 'match'));
end
Out = Col3(:,2)

This works for both when I tested it, amazingly enough!

(I still have much to learn about regexp myself.)

[Stephen23]

Another easy solution: '\d+\.?\d*'

[Star Strider]

Noted. Thank you.

Answer 2

[Andrei Bobrov]

In R2016b:

>> weather_strings = string({'UTC,2140991,49.0'
                           'UTC,2140992,49.1'
                           'UTC,2140993,49.1'})
weather_strings = 
  3x1 string array
    "UTC,2140991,49.0"
    "UTC,2140992,49.1"
    "UTC,2140993,49.1"
>> str2double(regexp(C,'(\d+\.)?\d+$','match','once'))
ans =
           49
         49.1
         49.1
>>