parf_predict.m

function [result, result2d] = parf_predict(data, model, diarypath, varargin)
    %
    % This script was written by Robert Weyres in 2015 and adapted for public usage in 2022.
    % A fork of randomforest-matlab used in this script can be found at https://github.com/jrderuiter/randomforest-matlab.
    %
    % The purpose of this function is to run prediction on multiple threads to save time.
    %
    % The script takes the following input parameters
    %
    % data - the data set to run the prediction on
    % model - the trained model
    % diarypath - the path to the diary used to reconstruct the order after prediction
    % varargin - num_threads, the number of threads to predict with - if not specified, num_threads is set to 25
    %
    % The script is licensed under the MIT license
    %
    % MIT License
    %
    % Copyright (c) 2015; 2022 Robert Weyres
    %
    % Permission is hereby granted, free of charge, to any person obtaining a copy
    % of this software and associated documentation files (the "Software"), to deal
    % in the Software without restriction, including without limitation the rights
    % to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
    % copies of the Software, and to permit persons to whom the Software is
    % furnished to do so, subject to the following conditions:
    %
    % The above copyright notice and this permission notice shall be included in all
    % copies or substantial portions of the Software.
    %
    % THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
    % IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
    % FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
    % AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
    % LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
    % OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
    % SOFTWARE.

    switch nargin
        case 3
            num_threads = 25

        case 4
            num_threads = varargin{1}

        otherwise
            disp('Error: Please input data, model and diarypath.')
            end
    end

    num_threads = 25; % Number of threads to be used
    delete(gcp('nocreate'))% Delete any existing parallel pools

    if size(gcp('nocreate'), 1) == 0% Create parpool with number of threads.
        tic
        poolobj = parpool('local', num_threads);
        toc
    end

    disp('Predicting...');
    % Parallel Prediction start
    disp('Splitting Input for Parallel Prediction...');
    result = [];

    for i = 1:num_threads - 1% Measure size of the source matrix and calculate sizes for splits.

        if ~exist('l', 'var')
            l{i, 1} = floor(size(data, 1) / num_threads - 1);
        else
            l{i, 1} = l{i - 1, 1} + floor(size(data, 1) / num_threads - 1);
        end

    end

    l{end + 1, 1} = size(data, 1) - l{end, 1};

    for i = 1:num_threads% Split the source matrix into to the previously calculated sizes.

        if ~exist('data_split', 'var')
            data_split{i, 1} = data(1:l{i}, :);
        elseif l{i} - l{i - 1} <= 0
            data_split{i, 1} = data(1 + l{i - 1}:end, :);
        else
            data_split{i, 1} = data(1 + l{i - 1}:l{i}, :);
        end

    end

    clear l

    if size(gcp('nocreate'), 1) == 0% Open parallel pool to perform the prediction on the chosen number of threads.
        tic
        poolobj = parpool('local', num_threads);
        toc
    end

    disp('Predicting...');
    diary diarypath% Enable diary to log the order in which the threads are finished. Needed for reconstruction of the result.
    % Parallel Prediction start
    parpredstart = tic;

    parfor i = 1:num_threads
        l2{i, 1} = classRF_predict(cell2mat(data_split(i, 1)), model, i); % Perform parallel prediction on the previously created splits.
        fprintf('%d\n', i);
    end

    parpredicttime = toc(parpredstart);
    % Parallel Prediction end
    diary off% Disable diary and import the created file into a matrix.
    fid = fopen(diarypath, 'r');
    formatSpec='%d';
    ord = fscanf(fid, formatSpec);
    fclose(fid);
    delete(diarypath);

    for i = 1:num_threads% Align the calculated predictions with the information from diary.
        l2{i, 2} = ord(i, 1);
        l2{ord(i, 1), 3} = l2{i, 1};
    end

    for i = 1:num_threads% Sort the prediction results.
        l3{i, 2} = ord(i, 1);
        l3{i, 1} = l2{ord(i, 1), 3};
    end

    clear ord
    clear l2
    % Resultat ohne Parpool korrekt zusammenfügen					    % Join the results into a single matrix.
    for i = 1:num_threads
        result = [result; l3{i, 1}];
    end

    clear l3
    disp('Finalising...');
    result2d = reshape(result2d, size(source, 1), size(source, 2));
    disp('Done.');
    clear data_split
end