Why doesn't the figure show the text and fitting line?

Text in the left corner and fitting line is missing from my fiures, please make correction to my code:

% Define heights, FOVs, and SNR values to test
heights = [1000, 2000, 3000, 4000];
fovs = [0.2, 0.5, 1, 2, 5, 10];
snr_values = [0, 25, 50, 75, 100];

% Function to calculate performance metrics
calculate_r_squared = @(x, y) 1 - sum((y - x).^2) / sum((y - mean(y)).^2);
calculate_rmse = @(x, y) sqrt(mean((y - x).^2));
calculate_mape = @(x, y) mean(abs((y - x) ./ y)) * 100;
calculate_mae = @(x, y) mean(abs(y - x));
calculate_made = @(x, y) mean(abs(y - mean(x)));

% Initialize arrays to store performance metrics
performance_metrics = struct();

% Loop over height values
for h = heights
    % Filter the data for the current height
    idx = (lookup_table(:, 1) == h);
    data_filtered = lookup_table(idx, :);

    % Initialize arrays to store performance metrics for each FOV and SNR value
    performance_metrics(h).r_squared_r = zeros(length(fovs), length(snr_values));
    performance_metrics(h).rmse_r = zeros(length(fovs), length(snr_values));
    performance_metrics(h).mape_r = zeros(length(fovs), length(snr_values));
    performance_metrics(h).mae_r = zeros(length(fovs), length(snr_values));
    performance_metrics(h).made_r = zeros(length(fovs), length(snr_values));
    performance_metrics(h).r_squared_a = zeros(length(fovs), length(snr_values));
    performance_metrics(h).rmse_a = zeros(length(fovs), length(snr_values));
    performance_metrics(h).mape_a = zeros(length(fovs), length(snr_values));
    performance_metrics(h).mae_a = zeros(length(fovs), length(snr_values));
    performance_metrics(h).made_a = zeros(length(fovs), length(snr_values));

    % Plot optimal_r_input vs. optimal_r_interp
    figure;
    hold on;
    colors = jet(length(fovs) * length(snr_values));
    c_idx = 1;
    for fov_idx = 1:length(fovs)
        for snr_idx = 1:length(snr_values)
            fov = fovs(fov_idx);
            snr = snr_values(snr_idx);

            % Filter data for the current FOV and SNR
            idx_fov_snr = (data_filtered(:, 2) == fov) & (data_filtered(:, 3) == snr);
            optimal_r_input = data_filtered(idx_fov_snr, 4);
            optimal_r_interp = data_filtered(idx_fov_snr, 5);

            % Scatter plot
            if ~isempty(optimal_r_input)
                scatter(optimal_r_input, optimal_r_interp, 50, colors(c_idx, :), 'filled');

                % Fit and plot linear regression line if there is sufficient data
                if length(optimal_r_input) > 1
                    model_r = fitlm(optimal_r_input, optimal_r_interp);
                    plot(model_r.Variables.x1, model_r.Fitted, 'Color', colors(c_idx, :), 'LineWidth', 2);

                    % Calculate additional performance metrics
                    r_squared_r = model_r.Rsquared.Ordinary;
                    rmse_r = calculate_rmse(optimal_r_input, optimal_r_interp);
                    mape_r = calculate_mape(optimal_r_input, optimal_r_interp);
                    mae_r = calculate_mae(optimal_r_input, optimal_r_interp);
                    made_r = calculate_made(optimal_r_input, optimal_r_interp);

                    % Store the performance metrics for this FOV and SNR value
                    performance_metrics(h).r_squared_r(fov_idx, snr_idx) = r_squared_r;
                    performance_metrics(h).rmse_r(fov_idx, snr_idx) = rmse_r;
                    performance_metrics(h).mape_r(fov_idx, snr_idx) = mape_r;
                    performance_metrics(h).mae_r(fov_idx, snr_idx) = mae_r;
                    performance_metrics(h).made_r(fov_idx, snr_idx) = made_r;

                    % Display text with performance metrics
                    text(mean(optimal_r_input), mean(optimal_r_interp), ...
                        {['SNR = ', num2str(snr), ' dB'], ...
                         ['R^2 = ', num2str(r_squared_r)], ...
                         ['RMSE = ', num2str(rmse_r)], ...
                         ['MAPE = ', num2str(mape_r), '%'], ...
                         ['MAE = ', num2str(mae_r)], ...
                         ['MADE = ', num2str(made_r)]}, ...
                        'FontSize', 10, 'Color', colors(c_idx, :));
                end
            end
            c_idx = c_idx + 1;
        end
    end

    xlabel('Optimal R_{e} (\mum)');
    ylabel('Optimal R_{e} interp (\mum)');
    title(['Plot of optimal R_{e} and optimal R_{e} interp for Height = ', num2str(h)]);
    grid on;
    hold off;

    % Plot optimal_a_input vs. optimal_a_interp
    figure;
    hold on;
    c_idx = 1;
    for fov_idx = 1:length(fovs)
        for snr_idx = 1:length(snr_values)
            fov = fovs(fov_idx);
            snr = snr_values(snr_idx);

            % Filter data for the current FOV and SNR
            idx_fov_snr = (data_filtered(:, 2) == fov) & (data_filtered(:, 3) == snr);
            optimal_a_input = data_filtered(idx_fov_snr, 6);
            optimal_a_interp = data_filtered(idx_fov_snr, 7);

            % Scatter plot
            if ~isempty(optimal_a_input)
                scatter(optimal_a_input, optimal_a_interp, 50, colors(c_idx, :), 'filled');

                % Fit and plot linear regression line if there is sufficient data
                if length(optimal_a_input) > 1
                    model_a = fitlm(optimal_a_input, optimal_a_interp);
                    plot(model_a.Variables.x1, model_a.Fitted, 'Color', colors(c_idx, :), 'LineWidth', 2);

                    % Calculate additional performance metrics
                    r_squared_a = model_a.Rsquared.Ordinary;
                    rmse_a = calculate_rmse(optimal_a_input, optimal_a_interp);
                    mape_a = calculate_mape(optimal_a_input, optimal_a_interp);
                    mae_a = calculate_mae(optimal_a_input, optimal_a_interp);
                    made_a = calculate_made(optimal_a_input, optimal_a_interp);

                    % Store the performance metrics for this FOV and SNR value
                    performance_metrics(h).r_squared_a(fov_idx, snr_idx) = r_squared_a;
                    performance_metrics(h).rmse_a(fov_idx, snr_idx) = rmse_a;
                    performance_metrics(h).mape_a(fov_idx, snr_idx) = mape_a;
                    performance_metrics(h).mae_a(fov_idx, snr_idx) = mae_a;
                    performance_metrics(h).made_a(fov_idx, snr_idx) = made_a;

                    % Display text with performance metrics
                    text(mean(optimal_a_input), mean(optimal_a_interp), ...
                        {['SNR = ', num2str(snr), ' dB'], ...
                         ['R^2 = ', num2str(r_squared_a)], ...
                         ['RMSE = ', num2str(rmse_a)], ...
                         ['MAPE = ', num2str(mape_a), '%'], ...
                         ['MAE = ', num2str(mae_a)], ...
                         ['MADE = ', num2str(made_a)]}, ...
                        'FontSize', 10, 'Color', colors(c_idx, :));
                end
            end
            c_idx = c_idx + 1;
        end
    end

    xlabel('Optimal \alpha_{e} (m^{-1})');
    ylabel('Optimal \alpha_{e} interp (m^{-1})');
    title(['Plot of optimal \alpha_{e} vs optimal \alpha_{e} interp for Height = ', num2str(h)]);
    grid on;
    hold off;
end

[Walter Roberson]

            idx_fov_snr = (data_filtered(:, 2) == fov) & (data_filtered(:, 3) == snr);

It is probably uncommon for there to be more than one exact match, so if length(optimal_a_input) > 1 probably just never happens.

[Wiqas Ahmad]

Thanks. But how to resolve it?

[Walter Roberson]

You have defined your code to output several text strings each time that length(optimal_a_input) > 1 . You have defined your code with

            idx_fov_snr = (data_filtered(:, 2) == fov) & (data_filtered(:, 3) == snr);
            optimal_a_input = data_filtered(idx_fov_snr, 6);

which will result in length(optimal_a_input) > 1 only if there are multiple matches for idx_fov_snr

You have defined

    idx = (lookup_table(:, 1) == h);
    data_filtered = lookup_table(idx, :);

So it all gets back to lookup_table: are there places in lookup_table that match h and for which there are multiple matches on fov and snr ? You have not posted the content of lookup_table so we do not know.

Note: it is risky to use == comparisons between fovs = [0.2, 0.5, 1, 2, 5, 10] and other arrays. 0.2 has no exact finite binary representation (for the same mathematical reason that 1/3 has no exact finite decimal representation.) You run the risk that what is being compared against is almost but not exactly the same as the 0.200000000000000011102230246251565404236316680908203125 that will be used for the hard-coded 0.2

[Wiqas Ahmad]

You can see my lookup table here:

[Walter Roberson]

You essentially have an unrolled 3D array. Each combination of height, fov, and snr is unique. length(optimal_a_input) will never be greater than 1

[Wiqas Ahmad]

Could you please be more specific Sir? I'm not getting the desired output

[Walter Roberson]

It is not possible for you to get the desired output with that code.

Your input table is such that it goes through all possible combinations of height, fov, and snr, with each combination appearing exactly once. But you only emit the text() if there is more than one match for a combination of height, fov, and snr -- which is something that never occurs in your table.

If you examine your table, you see that it holds the height constant for a while. Within height, it holds fov constant for a while, iterating over all snr. Then it moves on to the next fov within the height, and lists all snr. Then it moves on to the next fov, and so on. When all of the combinations of fov and snr have been run through, it moves on to the next height. At no point is there a duplicate combination of height, fov, and snr.

Answer 1

[arushi]

Hi Wiqas,

The code seems to be correct.To ensure that the text annotations and fitting lines are properly displayed on your figures, please consider the following adjustments and suggestions:

Ensure Text is Visible:

• Make sure the text is positioned within the plot limits. If the text is outside the axes limits, it won't be visible. Use the xlim and ylim functions to adjust the axes limits if necessary.

Adjust Text Position:

• Position the text based on the data range to ensure it is visible. You can use relative positions or fixed offsets to ensure the text doesn't overlap with data points.

Ensure Fitting Line is Plotted:

• Verify that the fitlm function is being called with sufficient data points. If there are too few points, the fitting line may not be plotted.

Hope this helps.