dividing a table into an array of subtables

Hello i have an array which look like in the photo bellow. every contactID has one or many activities(which are listed on the next rows under an empty contact id).

for example in line 15's contact id we have one activity and the next activity is at line 16.

how can i make an array of "contact_id"'s and in every cell we will have a sub array which adds all the activities if the following empty "contact id cell" as one array for this contact id?

thanks

mark |

Answer 1

[Robert]

In general, your task will be easier if you start by filling in those blank IDs. I will use data to refer to your table.

%%Fill in blanks (assumes first row is not blank)
for ii = 1:height(data)
  ID = data.ContactID{ii};
  if ~isempty(ID) % First occurance
      prev = ID;
  else % Not first occurance
      data.ContactID{ii} = prev;
  end
end

Then you can gather the IDs and select table subsets by common ID

[ID, ~, index] = unique(data.ContactID);
grouped = table();
grouped.ContactID = ID;
grouped.ActivityData = cell(height(grouped), 1);
for ii = 1:length(ID)
  grouped.ActivityData{ii} = data(index == ii, 2:end);
end

You could also do this in a containers.Map object instead of a table.

grouped = containers.Map();
for ID = unique(data.ContactID)'
  grouped(ID{1}) = data(strcmp(data.ContactID, ID{1}), 2:end);
end

[fima v]

Hello,my file is xls file which i imported as shown in the photo bellow,how can i change the import so it will match the 'data' variable type which you presented?

i have attache the xls file.

Thanks

[Guillaume]

@Robert,

Your first part (filling the blank) can be replaced by just one line using fillmissing:

data = fillmissing(data, 'previous', 'DataVariables', 1)

The second part can also be replaced by just one line, using rowfun. Granted, it's a bit obscure:

 grouped = rowfun(@(varargin) {table(varargin{:}, 'VariableNames', data.Properties.VariableNames(2:end))}, data, 'GroupingVariables', 1, 'InputVariables', 2:width(data))

@fima v,

Use readtable instead of the import tool

[Robert]

@Guillaume

Thanks for mentioning fillmissing; very cool! I am using R2016a most of the time, which is one release too old for fillmissing.

[fima v]

Hello ,how its possible keep only the latest activity which has the highest date among them , for each contact ?

Thanks

[Robert]

You should take a look at the second output of unique . It gives you the index of the first occurrence of each unique value. Specify that you want the 'last' occurrence and or sort your data by date as needed.