%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
% This script is used to merge policy data from different periods and calculate
% the corresponding averages as reported in:                                  
%                                                                              
% Daniel Nohrstedt, Maurizio Mazzoleni, Charles F. Parker, and Giuliano Di
% Baldassarre. Policy change after natural hazard
% events, Nature Communication
%                                                                              
% This script was prepared by Maurizio Mazzoleni                 
% Uppsala University, 2020                                                  
% mail to: maurizio.mazzoleni@geo.uu.se 
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%


%%Join period 2007-2009 with period 2009-2011 as they have different #countries
%Merge countries
count_p1_2=[querybuilder20072009;querybuilder20092011];
%Merge PFA
hfa_period1_2(:,1)=[nanmean(HFA_2007_2009(:,1:4),2);nanmean(HFA_2009_2011(:,1:4),2)]; %priority for action 1
hfa_period1_2(:,2)=[nanmean(HFA_2007_2009(:,5:8),2);nanmean(HFA_2009_2011(:,5:8),2)]; %PA 2
hfa_period1_2(:,3)=[nanmean(HFA_2007_2009(:,9:12),2);nanmean(HFA_2009_2011(:,9:12),2)]; %PA 3
hfa_period1_2(:,4)=[nanmean(HFA_2007_2009(:,13:18),2);nanmean(HFA_2009_2011(:,13:18),2)]; %PA 4
hfa_period1_2(:,5)=[nanmean(HFA_2007_2009(:,19:22),2);nanmean(HFA_2009_2011(:,19:22),2)];  %PA 5
%Sort
count_p1_1=sortrows(count_p1_2,1);
hfa_period1_sort1=cat(2,count_p1_2,num2cell(hfa_period1_2));
hfa_period1_sort2=sortrows(hfa_period1_sort1,1);
hfa_period1_3=hfa_period1_sort2(:,3:end);
hfa_period1_1=cell2mat(hfa_period1_3);
%Average calculation
for i=1:size(count_p1_1,1)
    if i==1
        count_p1{1,1}=count_p1_1{i,1};
        count_p1{1,2}=count_p1_1{i,2};
        hfa_period1(1,:)=(hfa_period1_1(i,:)+hfa_period1_1(i+1,:))/2;
    elseif i>1 && i<size(count_p1_1,1)
        if strcmp(count_p1_1{i-1,2},count_p1_1{i,2})==0 && strcmp(count_p1_1{i,2},count_p1_1{i+1,2})==0
            count_p1{i,1}=count_p1_1{i,1};
            count_p1{i,2}=count_p1_1{i,2};
            hfa_period1(i,:)=hfa_period1_1(i,:);
        elseif strcmp(count_p1_1{i-1,2},count_p1_1{i,2})==0 && strcmp(count_p1_1{i,2},count_p1_1{i+1,2})==1
            count_p1{i,1}=count_p1_1{i,1};
            count_p1{i,2}=count_p1_1{i,2};
            hfa_period1(i,:)=(hfa_period1_1(i,:)+hfa_period1_1(i+1,:))/2;
        elseif strcmp(count_p1_1{i-1,2},count_p1_1{i,2})==1 && strcmp(count_p1_1{i,2},count_p1_1{i+1,2})==0
            count_p1{i,1}=0;
            count_p1{i,2}=0;
            hfa_period1(i,:)=NaN;
        end
    elseif i==size(count_p1_1,1)
        if strcmp(count_p1_1{i-1,2},count_p1_1{i,2})==0
            count_p1{i,1}=count_p1_1{i,1};
            count_p1{i,2}=count_p1_1{i,2};
            hfa_period1(i,:)=hfa_period1_1(i,:);
        elseif strcmp(count_p1_1{i-1,2},count_p1_1{i,2})==1
            count_p1{i,1}=0;
            count_p1{i,2}=0;
            hfa_period1(i,:)=NaN;
        end
    end    
end
points1=count_p1;
HFAChange_mean_period1=hfa_period1;
C1 = points1(cellfun(@(x) ~isequal(x, 0), points1));
countryChange1 = reshape(C1, size(C1,1)/2, 2);
HFAChange_mean_period1(any(isnan(HFAChange_mean_period1), 2), :) = [];


%%Join period 2011-2013 with period 2013-2015 as they have different #countries
%Merge countries
count_p2_2=[querybuilder20112013;querybuilder20132015];
%Merge PFA
hfa_period2_2(:,1)=[nanmean(HFA_2011_2013(:,1:4),2);nanmean(HFA_2013_2015(:,1:4),2)]; %priority for action 1
hfa_period2_2(:,2)=[nanmean(HFA_2011_2013(:,5:8),2);nanmean(HFA_2013_2015(:,5:8),2)]; %PA 2
hfa_period2_2(:,3)=[nanmean(HFA_2011_2013(:,9:12),2);nanmean(HFA_2013_2015(:,9:12),2)]; %PA 3
hfa_period2_2(:,4)=[nanmean(HFA_2011_2013(:,13:18),2);nanmean(HFA_2013_2015(:,13:18),2)]; %PA 4
hfa_period2_2(:,5)=[nanmean(HFA_2011_2013(:,19:22),2);nanmean(HFA_2013_2015(:,19:22),2)];  %PA 5
%Sort
count_p2_1=sortrows(count_p2_2,1);
hfa_period2_sort1=cat(2,count_p2_2,num2cell(hfa_period2_2));
hfa_period2_sort2=sortrows(hfa_period2_sort1,1);
hfa_period2_3=hfa_period2_sort2(:,3:end);
hfa_period2_1=cell2mat(hfa_period2_3);
%Average calculation
for i=1:size(count_p2_1,1)
    if i==1
        count_p2{1,1}=count_p2_1{i,1};
        count_p2{1,2}=count_p2_1{i,2};
        hfa_period2(1,:)=(hfa_period2_1(i,:)+hfa_period2_1(i+1,:))/2;
    elseif i>1 && i<size(count_p2_1,1)
        if strcmp(count_p2_1{i-1,2},count_p2_1{i,2})==0 && strcmp(count_p2_1{i,2},count_p2_1{i+1,2})==0
            count_p2{i,1}=count_p2_1{i,1};
            count_p2{i,2}=count_p2_1{i,2};
            hfa_period2(i,:)=hfa_period2_1(i,:);
        elseif strcmp(count_p2_1{i-1,2},count_p2_1{i,2})==0 && strcmp(count_p2_1{i,2},count_p2_1{i+1,2})==1
            count_p2{i,1}=count_p2_1{i,1};
            count_p2{i,2}=count_p2_1{i,2};
            hfa_period2(i,:)=(hfa_period2_1(i,:)+hfa_period2_1(i+1,:))/2;
        elseif strcmp(count_p2_1{i-1,2},count_p2_1{i,2})==1 && strcmp(count_p2_1{i,2},count_p2_1{i+1,2})==0
            count_p2{i,1}=0;
            count_p2{i,2}=0;
            hfa_period2(i,:)=NaN;
        end
    elseif i==size(count_p2_1,1)
        if strcmp(count_p2_1{i-1,2},count_p2_1{i,2})==0
            count_p2{i,1}=count_p2_1{i,1};
            count_p2{i,2}=count_p2_1{i,2};
            hfa_period2(i,:)=hfa_period2_1(i,:);
        elseif strcmp(count_p2_1{i-1,2},count_p2_1{i,2})==1
            count_p2{i,1}=0;
            count_p2{i,2}=0;
            hfa_period2(i,:)=NaN;
        end
    end    
end
points2=count_p2;
HFAChange_mean_period2=hfa_period2;
C2 = points2(cellfun(@(x) ~isequal(x, 0), points2));
countryChange2 = reshape(C2, size(C2,1)/2, 2);
HFAChange_mean_period2(any(isnan(HFAChange_mean_period2), 2), :) = [];


%%Estimate difference HFA 2011-2015 and 2007-2011
count_p{1}=countryChange1;
count_p{2}=countryChange2;
hfa_period{1}=HFAChange_mean_period1;
hfa_period{2}=HFAChange_mean_period2;    
sizes3=cellfun(@length,count_p,'uni',false); %takes the number of countries per period (
[~,c3]=find(min(cell2mat(sizes3))==cell2mat(sizes3)); %selects which period has the smallest nr countries this number will be used for the loop below)
[~,m3]=find(max(cell2mat(sizes3))==cell2mat(sizes3)); %selects which period has the largest nr countries
d=1;
for ii=1:sizes3{c3}        
    temp_country=count_p{c3}{ii,2};  %takes the first country from the period with less number of countries
    temp_namecountry=count_p{c3}{ii,1};
    [row,~]=find(strcmp(count_p{m3}, temp_country)==1);  %finds if that country (temp_country) is present in the other period with more countries
    if ~isempty(row)
        if c3==1 %if the period with less countries is the first period
            Afirst=ii;  %then the row number is coinciding with the loop length, thus row number for this vector is -->'pais'
            Asecond=row;
        elseif c3==2  %if the period with less countries is the second period
            Afirst= row;
            Asecond=ii;
        end
        countryChange{d,1}=temp_country;  %writes cell with countries present in both periods, and thus policy change is possible to calculate
        countryChange{d,2}=temp_namecountry;
        if Norm_PFAchange==0
            HFAChange(d,:)=hfa_period{2}(Asecond,:)-hfa_period{1}(Afirst,:);
        elseif Norm_PFAchange==1
            HFAChange(d,:)=(hfa_period{2}(Asecond,:)-hfa_period{1}(Afirst,:))./hfa_period{1}(Afirst,:);
        end
        d=d+1;
    end
end