SAMI/MissingNodes_SrcCode2014/PrepareAttributes.m

% Map attribute to categories
% Use 0 for null/no value, 1 for unknown/private value (if exist) and then real values
function [outAttributes, attUpperRange] = PrepareAttributes(dataFilePath, dataFileName, numNodes, expectedAttrCols, selectedAttr, debug, debugPath)

outNoneValue = 0;
inNoneValue = -1;  

countryCol = 1;
countryCompress = 1;
countryOffset = 129; % min value is -128 => real value starting from 1
%countryUnknown = -999;

if nargin >= 3
    if nargin < 5
        selectedAttr = ones(1,expectedAttrCols);
    end
    if nargin < 6
        debug = 0;
    end
    if nargin < 7
        debugPath = 'debug/';
    end
    
    outFullPath = strcat(dataFilePath, debugPath);
    outFullName = strcat(outFullPath,dataFileName);

    if isdir(outFullPath) == 0 && debug == 1 
        mkdir(outFullPath);
    end
    
    attributes = LoadAsciiAttributes(dataFilePath, dataFileName, numNodes, debug, debugPath);
    m = size(attributes,1); % num nodes/lines
    n = size(attributes,2); % num attributes/cols
    outAttributes = zeros(m,nnz(selectedAttr)); % num cols according to num selectedAttr
    
    if n == expectedAttrCols
        for line=1:m
            column = 0;
            if selectedAttr(countryCol) == 1
                % country (col=1)
                column = column +1;
                value =  attributes(line, countryCol)+countryOffset;
                if (value < 0)
                    outAttributes(line, column) = outNoneValue;
                else
                    outAttributes(line, column) = value;
                end
            end

            % games/groups (all other columns)
            for inx = 2 : expectedAttrCols
                if selectedAttr(inx) == 1
                    column = column +1;
                    value =  attributes(line,inx);
                    if value == inNoneValue
                        outAttributes(line, column) = outNoneValue;
                    else % no threshold, i.e. binary value
                        outAttributes(line, column) = 1;
                    end
                end
            end
        end
        
        % remove empty country indexes and shift codes
        if countryCompress && selectedAttr(countryCol) == 1
            new_values = outAttributes(:, countryCol);
            maxCountry = max(new_values);
            for i=1:maxCountry
                indices = find(new_values==i);
                if size(indices,1) == 0
                    indices = find(new_values>i);
                    if size(indices,1) == 0
                        break;
                    else
                        new_values(indices) = new_values(indices) -1;
                    end
                end
            end
            %maxCountry = max(new_values);
            outAttributes(:, countryCol)= new_values;
        end

        attUpperRange = max(outAttributes);
        %attLowRange = min(attributes);
        
        if debug == 1
            outFullName = sprintf('%s.att2%d', outFullName, numThreshold);
            save(strcat(outFullName,'.mat'), 'outAttributes');
            SaveIntMatrixToFile(strcat(outFullName,'.txt'), outAttributes);
        end
    else
        fprintf('PrepareAttributes - Invalid attributes count: expecting %d, got %d\n',expectedAttrCols,n);
    end
else
    fprintf('PrepareAttributes - Invalid parameters. expecting: dataFilePath, rawAtts, debug\n');
end