Hex to Signed Int conversion of 16bit values in a Table with data type 'cell array of character vectors'

I have a table as below having huge number of rows.

Time                Identifier           AC1     AC2     AC3     AC4
00:06:40.23             "300"           "6500"  "6900"  "6D00"  "7500" 
00:06:40.25             "100"           "B7FF"  "E5FF"  "7D10"  "0100"
  ...                    ...              ...     ...     ...     ...

The columns AC1 to AC4 are in 16bit 2's complement representation.

Need to convert all the values of those 4 columns in to signed decimal vlaue. The table shld be intact.

Any help is appreciated. Also, would like to know what would be a convenient way to have these kind of data other than in table, so that would be helpful for further processing. The data being sensor values wrt time.

[Walter Roberson]

Please show

class(YourTable.AC2)

replacing YourTable with the name of your table.

[manoj hanu]

It returned 'cell'

Answer 1

[Walter Roberson]

temp = uint16(arrayfun(@(S) sscanf(S, '%x'), YourTable{:,3:6}));
output = reshape( typecast(temp(:), 'int16'), size(temp));

What kind of processing do you want to do on the entries? Possibly using a timetable() would be appropriate.

[manoj hanu]

.Thank you for the answer. There is a type mismatch I guess.

Error using sscanf
First argument must be a string scalar or character vector.

Error in @(S)sscanf(S,'%x')

Those are the reading from an Accelerometer. So probably I might need to use it further as an input to some filters for sensor integration and, also for plotting.

[Walter Roberson]

Try

temp = uint16(cellfun(@(S) sscanf(S, '%x'), YourTable{:,3:6}));
output = reshape( typecast(temp(:), 'int16'), size(temp));

[manoj hanu]

Sorry for streching this one long. That works But, completely becomes a new matrix. I would like to have the table intact with the column names.

[Walter Roberson]

NewTable = [YourTable(:,1:2), array2table(output)];
NewTable.Properties.VariableNames(3:end) = Table.Properties.VariableNames(3:end);

Answer 2

[Guillaume]

It returned 'cell'

Assuming that the "" in your example are actual double quotes and not single quotes, then you've not created your tables very well. The type of these columns should be string. A cell array of scalar string is a big waste of memory.

Anyway, it doesn't matter since we're converting it to numeric.

Assuming, R2018b or later:

yourtable = convertvars(yourtable, {'AC1', 'AC2', 'AC3', 'AC4'}, @(var) cellfun(@(s) sscanf(s, '%x'), var))

on earlier versions, it's a bit more complicated as you can't easily change the type of column. This should work, assuming the AC columns are the last 4:

orignames = yourtable.Properties.VariableNames(end-3:end);
yourtable = [yourtable(:, 1:end-4), ...
             varfun(@(var) cellfun(@(s) sscanf(s, '%x'), var), yourtable(:, end-3:end))];
yourtable.Properties.VariableNames(end-3:end) = orignames;  %varfun prepends Fun_ to the variable it processes

[manoj hanu]

The first one looks really helpful and would save a lot of time. But, unfortunately I am using 2018a. Thanks for the answer.