Predicting Next Value using fitrsvm and predict giving the constant output

1 次查看（过去 30 天）

Almost Artsy 2019-3-2

1
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/447833-predicting-next-value-using-fitrsvm-and-predict-giving-the-constant-output

Hello. I have trained a regression model using fitrsvm, then I use the predict function on my test data. The model performs really well. But, When I tried using "predict", it gives the same output array. Actually, I'm trying to predict the value of yts from 501 until 507 and compare the output with the real value. Please Help me.

%Data Train and Data Test
xtr = [nomor(1:500)];
ytr = [y(1:500)'];
xts = [nomor(501:507)];
yts = [y(501:507)']
mdl2 = fitrsvm(xtr,ytr,'KernelFunction','gaussian','epsilon',0.1,'Solver','ISDA');
conv = mdl2.ConvergenceInfo.Converged;
iter = mdl2.NumIterations;
yfit = predict(mdl2,xts)
% % % % % % These are the outputs % % % % % %
yfit =
   1.0e+03 *
    7.7353
    7.6361
    7.6312
    7.6312
    7.6312
    7.6312
    7.6312