arrays - Apply function to table conditioned on several variables in Matlab -

i have following 2 tables, data , members:

data = table(sort(repmat(datenum(2001,1:5,1).',4,1)),repmat(('a':'d').',5,1),repmat((201:204).',5,1),'variablenames',{'date','id','price'});  data =          date       id    price     __________    __    _____      7.3085e+05        201       7.3085e+05    b     202       7.3085e+05    c     203       7.3085e+05    d     204       7.3088e+05        201       7.3088e+05    b     202       7.3088e+05    c     203       7.3088e+05    d     204       7.3091e+05        201       7.3091e+05    b     202       7.3091e+05    c     203       7.3091e+05    d     204       7.3094e+05        201       7.3094e+05    b     202       7.3094e+05    c     203       7.3094e+05    d     204       7.3097e+05        201       7.3097e+05    b     202       7.3097e+05    c     203       7.3097e+05    d     204  members = table(datenum(2001,1:5,1).',{cell2table({'b','c'});table({'a'});cell2table({'b','d'});cell2table({'a','c'});cell2table({'a','c'})},'variablenames',{'date','memberid'});   members =          date        memberid       __________    ___________      7.3085e+05    [1x2 table]     7.3088e+05    [1x1 table]     7.3091e+05    [1x2 table]     7.3094e+05    [1x2 table]     7.3097e+05    [1x2 table] 

now, apply function, mean, price variable each ´date´ , ids in memberid on each date. thus, first date, 7.3085e+05, mean of id 2 , 3 - i.e. (202+203)/2 - etc.

i able slow for loop (my actual tables larger). thinking should possible using varfun or similar, can't work. ideas?

update

changed ids text, since how real data is.

approach 1

%// find offsets used linear indexing elements each group off1 = [0 ; find(diff(data.id)<0)]  %// store prices , member ids numeric arrays prc = data.price mem_id = members.memberid  %// convert each memberid table arrays; add corresponding %// linear indexing offsets , index prc , mean values out = arrayfun(@(n) mean(prc(off1(n)+ char(table2array(mem_id{n}))-'a'+1 )),...                                                             1:numel(mem_id)) 

approach 2

%// find offsets used linear indexing elements each group off1 = [0 ; find(diff(data.id)<0)]  %// store prices , member ids numeric arrays prc = data.price mem_id = members.memberid  %// storage output out = zeros(1,numel(mem_id))  %// widths(lengths) of each table in memberid , unique lengths lens = arrayfun(@(n) width(mem_id{n}),1:numel(mem_id)) unqlens = unique(lens,'stable')  %// now, have groups of memberid tables uniform %// widths/lengths, can use table2array on tables single %// call within each group , beneficial in performance iter = 1:numel(unqlens)     mask = lens==unqlens(iter); %// mask each group     mem_id_iter = mem_id(mask); %// member ids     grp_ids = reshape(char(table2array(vertcat(mem_id_iter{:}))) - 'a' + 1,[],...                                                 unqlens(iter)); %// group ids     lin_idx = bsxfun(@plus,off1(mask),grp_ids); %// linear indices     out(mask) = mean(prc(lin_idx),2); %// index prc , mean values end