My one-sample t-test isn't working as I need it to.

16 次查看（过去 30 天）
Ruslan Avramenko
Ruslan Avramenko 2024-4-26
编辑： Voss 2024-4-30
Hello everyone. I need to calculate the p-value for each sample with different dimensions and compare them with the reference value, which is the largest sample at 100 mm. I need to perform a one-sample t-test with a significance level of 0.05. Here is my code. I don't know why it always outputs NaN for the 80 mm value. I would like to fix this.
My Code:
% Clear the workspace, command window, and close any open figures
clc;
clear all;
close all;
% Input data
data = [67.142880, 69.218820, 67.379050, 69.943850, 67.407470]; % Data values
rozmery = [100, 80, 60, 40, 20]; % Dimensions
reference_value = data(1); % Reference value
% Significance level
alpha = 0.05;
% Initialize an array to store p-values
p_values = zeros(1, length(data) - 1);
% Perform t-tests and calculate p-values for each subsequent data point
for i = 2:length(data)
% Perform a one-sample t-test
[~, p_values(i-1)] = ttest2(data(1:i-1), data(i), 'Alpha', alpha);
end
% Plot the graph
figure;
plot(rozmery(2:end), p_values, '-o', 'LineWidth', 2); % Plot dimensions against p-values
xlabel('Dimensions (mm)'); % X-axis label
ylabel('P-values'); % Y-axis label
title('P-values vs Dimensions'); % Title of the plot
grid on; % Turn on the grid
% Add labels to the plot
for i = 2:length(rozmery)
text(rozmery(i), p_values(i-1), sprintf('(%d, %f)', rozmery(i), p_values(i-1)),...
'VerticalAlignment', 'bottom', 'HorizontalAlignment', 'left');
% Add text annotations for each point on the plot
end
% Display the corresponding dimension values and p-values
for i = 2:length(rozmery)
fprintf('P-value for dimension %d mm: %f\n', rozmery(i), p_values(i-1));
end
% Find the point where the p-value falls below the significance level
index_rustu = find(p_values < alpha, 1);
% Display the point where the p-value falls below the significance level
if ~isempty(index_rustu)
hold on;
plot(rozmery(index_rustu + 1), p_values(index_rustu), 'r*', 'MarkerSize', 10);
hold off;
fprintf('Point where the p-value falls below the significance level: %d mm\n', rozmery(index_rustu + 1));
fprintf('P-value at this point: %f\n', p_values(index_rustu));
else
disp('No point found where the p-value falls below the significance level.');
end

请先登录，再回答此问题。

采纳的回答

arushi
arushi 2024-4-30
Hi Ruslan,
The code uses "ttest2" for a two-sample t-test, but you mentioned needing a one-sample t-test to compare each sample with the reference value. This discrepancy might be one of the reasons why you're encountering unexpected results, such as `NaN` for the 80 mm value. The `NaN` could arise because you're comparing a single value against another single value at some point, which is not valid for a two-sample t-test.
To fix this, you should use "ttest" instead of "ttest2". The "ttest" function is designed for one-sample t-tests or paired samples, which seems more aligned with your requirement to compare each sample against the reference value.
Please refer to the following documentation of one-sample and two-sample t-test for more information-
https://www.mathworks.com/help/stats/ttest2.html
https://www.mathworks.com/help/stats/ttest.html
Hope this helps.
  1 个评论
Ruslan Avramenko
Ruslan Avramenko 2024-4-30
Hi Arushi,
I've already tried using ttest instead of ttest2. The code gave me different results, but the problem remained the same. I didn't obtain results for 80 mm.
Here is my code:
for i = 2:length(data)
[~, p_values(i-1)] = ttest(data(1:i-1), reference_value, 'Alpha', alpha);
end
Thank you for help

请先登录，再进行评论。

更多回答（0 个）

类别

在 Help Center 和 File Exchange 中查找有关 Descriptive Statistics 的更多信息

标签

产品


版本

R2023a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by