SunkCostModelsAndData2022/Test_Range.m at main · adredish/SunkCostModelsAndData2022 · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
function SCout = Test_Range(nB, varargin)

calcAttritionBias = true;
calcPEV = true;
showFigures = true;

sigmaN = [0 2 3 5];
sigmaW = [0 0.25 0.5 1 3 5 8 10 20];
process_varargin(varargin);

SCout.sigmaN = sigmaN; nN = length(SCout.sigmaN);
SCout.sigmaW = sigmaW; nW = length(SCout.sigmaW);

[sN, sW] = ndgrid(SCout.sigmaN, SCout.sigmaW);
SCout.baseSlope = nan(nB, nN, nW);
SCout.sunkCost = nan(nB, nN, nW);
SCout.threshold = nan(nB, nN, nW);
SCout.temperature = nan(nB, nN, nW);
SCout.attritionBias = nan(nB, nN, nW);

SCout.slopes.xrange = [-30 30];
SCout.slopes.nBins = 20;
SCout.slopes.x = nan(nB, nN*nW, SCout.slopes.nBins);
SCout.slopes.mu = nan(nB, nN*nW, SCout.slopes.nBins);
SCout.slopes.se = nan(nB, nN*nW, SCout.slopes.nBins);

[~,SCout.N, SCout.W] = ndgrid(1:nB, SCout.sigmaN, SCout.sigmaW);

t = nan(nB, nN*nW,1);
for iB = 1:nB
    if showFigures
        hSC = figure; set(hSC, 'units', 'normalized','outerposition', [0.5 0 0.5 1]);
        hAB = figure; set(hAB, 'units', 'normalized','outerposition', [0.5 0 0.5 1]);
        hPV = figure; set(hPV, 'units', 'normalized','outerposition', [0.5 0 0.5 1]);
    end

    for iS = 1:(nN*nW)
        tic;
        fprintf('%d/%d: [%.2f, %.2f]: ', iB, nB, sW(iS), sN(iS));

        R = GenerateSimulation('sigmaN', sN(iS), 'sigmaW', sW(iS));

        fprintf(' pEarn ');
        pEarn = Calculate_pEarn(R);
        [S,C] = Calculate_SunkCostSlopes(pEarn,R);
        SCout.baseSlope(iB,iS) = S(1);
        SCout.sunkCost(iB,iS) = nansum(S(:,1) - C(:,1));

        if calcAttritionBias
            fprintf(' AttritionBias ');
            [~,SCout.attritionBias(iB,iS)] = Calculate_AttritionBias(R);
        end

        if calcPEV
            fprintf(' PEV ');
            PEV = Calculate_pEarnXvalue(R);
            slopes = Calculate_PEV_slopes(PEV, ...
                'nBins', SCout.slopes.nBins, 'xrange', SCout.slopes.xrange);
            SCout.slopes.x(iB,iS,:) = slopes.x;
            SCout.slopes.mu(iB,iS,:) = slopes.mu;
            SCout.slopes.se(iB,iS,:) = slopes.se;
        end

        fprintf(' probit ');
        b = glmfit(R.offer, R.isStay, 'binomial','link', 'probit');
        [SCout.threshold(iB,iS), SCout.temperature(iB,iS)] = unpackProbit(b);

        if sW(iS) < 1
            MSG = sprintf('\\sigma_W = %.2f \\sigma_N = %.0f', sW(iS), sN(iS));
        else
            MSG = sprintf('\\sigma_W = %.0f \\sigma_N = %.0f', sW(iS), sN(iS));
        end

        if showFigures
            fprintf(' drawing ');
            % SunkCosts
            figure(hSC);
            subplot(nW, nN, iS);
            Show_pEarn(pEarn, R, 'filled', false);
            legend off; title('');
            text(1,0.05,sprintf('%s: SC=%.2f, BS=%.2f', MSG, SCout.sunkCost(iB,iS), SCout.baseSlope(iB,iS)));
            drawnow;

            if calcAttritionBias
                % attritionbias
                figure(hAB); subplot(nW, nN, iS);
                [~,AB] = Show_AttritionBias(R);
                legend off; title('');
                text(1,5,sprintf('%s, AB = %.2f', MSG, AB), 'color', 'w');
                drawnow;
            end

            if calcPEV
                % pEarnXvalue
                figure(hPV); subplot(nW, nN, iS);
                Show_pEarnXvalue2(PEV); legend off
                text(-25, -0.2, MSG);
                drawnow;
            end
        end
        t(iB,iS) = toc;
        fprintf('%.0f sec (%.0f sec to go)\n', t(iB,iS), mean(t(:),'omitnan') * sum(isnan(t(:))));
    end
end

end