Trend lines with moving average

Question

Max1234 2023-2-26

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1919170-trend-lines-with-moving-average

回答： Steven Lord 2023-2-26

采纳的回答： Star Strider

Test.mat

Hi guys,

I have a data set with the following double values (Test.mat). When i plot the Data with scatter i get this figure:

I now need two types of trend lines. The first should be linear and the second should represent as many "curves" as possible. How can I achieve this with a moving average?

I need the trend line as a function so that I can insert it together with other trend lines into a new plot and compare them.

Many thanks for your help!

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

请先登录，再回答此问题。

Answer 1

Star Strider 2023-2-26

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1919170-trend-lines-with-moving-average#answer_1180215

在 MATLAB Online 中打开

It is unlikely that it is posssible to get any meanigful information from these unless they are first sorted.

LD = load(websave('Test','https://www.mathworks.com/matlabcentral/answers/uploaded_files/1307620/Test.mat'))

LD = struct with fields:

x_Test: [67×1 double] y_Test: [67×1 double]

x = LD.x_Test;

y = LD.y_Test;

xys = sortrows([x,y],1);

x = xys(:,1);

y = xys(:,2);

window = 3; % Window Over Which The Moving Average Is Computed

ymm = movmean(y,window);

DM = [x ones(size(x))];

B = DM \ y;

ylr = DM * B;

figure

plot(x, y, '.', 'DisplayName','Sorted Data')

hold on

plot(x, ymm, 'DisplayName','Moving Average')

plot(x, ylr, 'LineWidth',2, 'DisplayName','Linear Regression')

hold off

xlabel('X')

ylabel('Y')

grid

legend('Location','best')

The data needs to be sorted to compute the moving average (the movmean function). The regression does not care if the data are sorted or not, however the plot of the regression line very much cares.

.

0 个评论
显示 -2更早的评论隐藏 -2更早的评论

请先登录，再进行评论。

Answer 2

the cyclist 2023-2-26

0
链接

此回答的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/1919170-trend-lines-with-moving-average#answer_1180220

在 MATLAB Online 中打开

Test.mat

There are many ways to do the linear fit in MATLAB, including some in base MATLAB. Here is one way, using the fitlm function from the Statistics and Machine Learning Toolbox. I expect other folks will post answers using other functions.

load("Test.mat","x_Test","y_Test")

mdl = fitlm(x_Test,y_Test);

figure

hold on

scatter(x_Test,y_Test)

plot(sort(x_Test),predict(mdl,sort(x_Test)))

I'm unclear what you mean by the second thing you need, with "as many curves as possible". One could make a perfect fit through these points, with any number of curves (e.g. with a polynomial of extremely high order). You need to clarify what you mean here.

Also, I'm not sure what you mean by a moving average here. I know what a moving average is, but it's unclear how you want to include that along with a trend line.