importcsv_and_7.m

clc, clear all, close all
file=dir('android_remdup\*');
gpsData=cell(1,length(file)-2);

x=1; % survey period counter 
survey_interval = 14; % cahnge to 14 for phq9
coverday_threshod = 4;  % how many days need to be covered in one period
accth = 165; % threshold for accuracy
for n=3:length(file)
    file(n).name
    tmp=csvread(['android_remdup\',file(n).name],1,0);
    tmp=sortrows(tmp,4);
    tmp=[tmp ones(length(tmp),1).*-1];
    dep = tmp(1,8);
    
    wifitmp=csvread(['wifiprocess\',file(n).name],1,0);
    for i=2:length(wifitmp(:,1))  % disconnection events
        if wifitmp(i,3)==0
            wifitmp(i,1:2)=wifitmp(i-1,1:2);
        end
    end
    wifitmp(find(wifitmp(:,1)==-1),:)=[];  % no GPS matched
    wifitmp(find(wifitmp(:,2)==-1),:)=[];
    wifitmp=[zeros(length(wifitmp(:,1)),1) wifitmp(:,2) wifitmp(:,1) wifitmp(:,4) ones(length(wifitmp(:,1)),1).*tmp(1,5) ones(length(wifitmp(:,1)),1).*-1 ones(length(wifitmp(:,1)),1).*-1 ones(length(wifitmp(:,1)),1).*-1 ones(length(wifitmp(:,1)),1).*-1 wifitmp(:,3).*-2];
    
    gpsTime = datestr(wifitmp(:,4)/86400000 + datenum(1970,1,1));
	t = datetime(gpsTime, 'Format', 'dd/M/yy HH:mm:ss');
	t = datevec(t);
    tmp = [tmp;wifitmp];
    tmp=sortrows(tmp,4);
    
    actitmp_=[];
    actitmp=csvread(['actiprocess\',file(n).name],1,0);
    actitmp = actitmp(find(actitmp(:,3)<max(tmp(:,4))&actitmp(:,3)>min(tmp(:,4))),:);
    actitmp = actitmp(find(actitmp(:,4)>=3&actitmp(:,7)>50),:);%moving with accuracy>50 
    fill = zeros(length(actitmp(:,1)),1);
    actitmp_=[fill fill fill actitmp(:,3) fill fill fill fill actitmp(:,2) ones(length(actitmp(:,1)),1).*3];
    gpsTime = datestr(actitmp_(:,4)/86400000 + datenum(1970,1,1));
	t = datetime(gpsTime, 'Format', 'dd/M/yy HH:mm:ss');
	t = datevec(t);
    tmp = [tmp;actitmp_];
    tmp=sortrows(tmp,4);
    
    % update the survey date for wifi data
    unip = sortrows(unique(tmp(:,9)),1);
    unip(find(unip(:)==-1))=[];
    for i = 2:length(tmp)
        if tmp(i,6)==-1
            for j=1:length(unip)
                    if (tmp(i,4)<unip(j))&&(tmp(i,4)>unip(j)-survey_interval*1440*60*1000)
                        tmp(i,9)= unip(j); tmp(i,8)= dep;
                        break
                    end
            end
        end
    end
    
    % aplit data based on survey date
    unip = sortrows(unique(tmp(:,9)),1);
    unip(find(unip(:)==-1))=[];
    unip(find(unip(:)<1.4*10^12))=[];
    for i=1:length(unip)
        index = find(tmp(:,9)==unip(i));
        if length(index)>100
            gpsData{1,x}=[ones(length(index),1).*unip(i) tmp(index,[2:4])  zeros(length(index),1) tmp(index,5:7) zeros(length(index),1) tmp(index,[8 10])];
            gpsData{1,x}=gpsData{1,x}(find(gpsData{1,x}(:,8)<=accth),:); 
            gpsData{1,x}=sortrows(gpsData{1,x},4);
            x=x+1;
        end
    end
    file(n).name;
end 


%calculate week day
for i = 1:size(gpsData,2)
    temp = size(gpsData{1,i},2);
    % Get day of the week info
    wd = weekday(datestr(gpsData{1,i}(:,4)/86400000 + datenum(1970,1,1)));
    gpsData{1,i}(:,9) = wd;
end

% calculate number of days covered in one period
sdc=[]
for i = 1:size(gpsData,2)
    temp = gpsData{1,i};
    if length(temp)==0
        continue;
    end
    dc =1;
    for di=2:length(temp)
        if temp(di,9)~=temp(di-1,9)
            dc=dc+1;
        end
    end
    sdc = [sdc dc];
end
gpsData(find(sdc(1,:)<=coverday_threshod))=[];
id = cellfun('length',gpsData);
gpsData(id==0)=[]
save('gps_feature_and.mat','gpsData');