How does one display the fit line in this plot in this situation?

I have a timetable T with

T.date  T.vistors T.sales
----------------------------------------

I draw a plot

yyaxis left
plot(T.date, T.visitors)
hold on
yyaxis right
plot(T.date, T.sales)

Now I want to add a regression line of T.sales = a + b*T.visitors on this plot.

I looked at https://www.mathworks.com/matlabcentral/answers/64754-how-can-i-include-an-equation-in-a-scatter-plot-with-linear-regression-line

But this situation seems to be more complex. How to proceed?

Answer 1

[Sulaymon Eshkabilov]

T = readtable('https://www.mathworks.com/matlabcentral/answers/uploaded_files/717514/test.csv');

Warning: The DATETIME data was created using format 'MM/dd/uuuu' but also matched 'dd/MM/uuuu'.
To avoid ambiguity, supply a datetime format using SETVAROPTS, e.g.
opts = setvaropts(opts,varname,'InputFormat','MM/dd/uuuu');

 fit_model = polyfit(T.Visitor, T.Sales,1);
 TSales_Fit = polyval(fit_model,T.Visitor);
 yyaxis left
plot(T.Date, T.Visitor, 'ro', 'markerfacecolor', 'c', 'markersize', 13)
hold on
yyaxis right
plot(T.Date, T.Sales, 'bs', 'markerfacecolor', 'y', 'markersize', 9)
 hold on
 plot(0:numel(TSales_Fit)-1,TSales_Fit, 'k-', 'linewidth', 2)
 legend('Date vs. Visitors', 'Date vs. Sales', 'Fit model: Visitors vs. Sales', 'location', 'northwest') 
 grid on

[alpedhuez]

Make sense. Let me work on it.

Answer 2

[Sulaymon Eshkabilov]

You can use these additional steps to get the linear fit model and calculate its values, and then plot them.

 fit_model = polyfit(T.visitors,T.sales,1);
 TSales_Fit = polyval(fit_model,T.visitors);
 yyaxis left
plot(T.date, T.visitors)
hold on
yyaxis right
plot(T.date, T.sales)
 hold on
 plot(T.visitors, TSales_Fit)

Note that your set x -axis is T.date and what you are trying to plot on top of that is T.visitors (x-axis) vs. Tsales_Fit (y-axis). There is a mismatch. Thus, it is more appropriate to compute fit model: DD=datenum(T.date); TSales_Fit = a + b*(DD-DD(1)+1); and plot the followings:

 DD=datenum(T.date); 
 DD = DD-DD(1)+1;
 fit_model = polyfit(DD,T.sales,1);
 TSales_Fit = polyval(fit_model,DD);
 yyaxis left
plot(T.date, T.visitors)
hold on
yyaxis right
plot(T.date, T.sales)
 hold on
 plot(T.date, TSales_Fit)
 legend('Date vs. Visitors', 'Date vs. Sales', 'Fit model: Date vs. Sales', 'location', 'best')

[alpedhuez]

I am not sure. I am not looking to plot T.sales - a+ b*T.date. I want to plot T.sales = a+b*T.vistors and then I want to plot it on x-axis date and y-axis sales. Get sales on that date and then apply linear regerssion to estiamte sales.

[Image Analyst]

Please attach your table T in a .mat file with the paper clip icon if you need any more help. Otherwise we're just guessing.

[alpedhuez]

I have attached a simple csv file. Thank you.