How to import 'kfoldPredict' for old version matlab (2023a)

I want to use kfold with decision tree model. But kfoldpredict is for matlab 2024a.

Is there a possible way to import such function/ packages to matlab 2023?

I try save the class after open 'kfoldpredict' function, but it seems does not work.

Thank you for possible solutions.

Answer 1

[Ayush Modi]

Hi Evelyn,

"kfoldPredict" function was introduced in r2011a. You can use the function in r2023 as well.

However, "kfoldPredict" function is a part of "Statistics and Machine Learning Toolbox". So you would need the license to the r2023 version of the toolbox.

Refer to the following MathWorks documentation to learn more about the history of the function across various versions:

https://www.mathworks.com/help/stats/classreg.learning.partition.classificationpartitionedmodel.kfoldpredict.html?s_tid=doc_ta#:~:text=Parallel%20Computing%20Toolbox%E2%84%A2.-,Version%20History,-Introduced%20in%20R2011a

[evelyn]

thank you for your reply.

I already installed 'Statistics and Machine Learning Toolbox'. And still get warning 'Unrecognized function or variable 'kfoldPredict'.'

[evelyn]

thank you again!

I try the example in kfoldPredict

It seems work fine with ''kfoldPredict'', not sure whether it is because other part of my code.

[Steven Lord]

How exactly are you calling kfoldPredict? The specific input arguments with which you're calling it could determine which kfoldPredict method gets called (or whether MATLAB can call that method at all!)

[evelyn]

thank for your reply!

Here is my code, the data are randomlt generated(not the real one that I used).

dataNum =200;
featureNum=4;
features = randn(dataNum,featureNum);
response = randi([0 1], dataNum, 1);

% 5-fold
t = templateTree('MinLeafSize', 10);
rfModel = fitcensemble(features, response,'Method', 'Bag', 'Learners', t,'OptimizeHyperparameters','auto');

|===================================================================================================================================|
| Iter | Eval   | Objective   | Objective   | BestSoFar   | BestSoFar   |       Method | NumLearningC-|    LearnRate |  MinLeafSize |
|      | result |             | runtime     | (observed)  | (estim.)    |              | ycles        |              |              |
|===================================================================================================================================|
|    1 | Best   |        0.49 |      6.4539 |        0.49 |        0.49 |   AdaBoostM1 |          388 |       0.2419 |            1 |
|    2 | Accept |        0.52 |      2.2347 |        0.49 |     0.49119 |  GentleBoost |          131 |     0.059663 |            9 |
|    3 | Accept |        0.59 |      2.4104 |        0.49 |     0.49463 |     RUSBoost |          129 |     0.044567 |           55 |
|    4 | Best   |       0.485 |     0.55492 |       0.485 |     0.48514 |  GentleBoost |           33 |    0.0010033 |            2 |
|    5 | Accept |       0.515 |      0.6316 |       0.485 |     0.50832 |  GentleBoost |           35 |    0.0012075 |            1 |
|    6 | Best   |        0.45 |     0.70127 |        0.45 |     0.46756 |   AdaBoostM1 |           52 |    0.0010006 |           98 |
|    7 | Accept |        0.45 |     0.67801 |        0.45 |        0.45 |   AdaBoostM1 |           50 |    0.0011996 |           96 |
|    8 | Accept |        0.54 |     0.51508 |        0.45 |     0.44997 |   AdaBoostM1 |           36 |     0.028928 |           34 |
|    9 | Accept |        0.45 |     0.33061 |        0.45 |     0.44997 |  GentleBoost |           19 |     0.034428 |           98 |
|   10 | Accept |       0.455 |     0.31386 |        0.45 |     0.44994 |  GentleBoost |           18 |     0.019053 |           65 |
|   11 | Accept |        0.46 |     0.24792 |        0.45 |     0.44991 |  GentleBoost |           12 |      0.12908 |           33 |
|   12 | Accept |        0.49 |     0.24321 |        0.45 |     0.44992 |   AdaBoostM1 |           13 |    0.0010767 |            3 |
|   13 | Accept |        0.45 |     0.22136 |        0.45 |     0.44989 |  GentleBoost |           10 |     0.070155 |           83 |
|   14 | Accept |        0.46 |     0.21113 |        0.45 |     0.44998 |   AdaBoostM1 |           11 |    0.0017986 |            7 |
|   15 | Accept |       0.505 |       0.192 |        0.45 |     0.44995 |   AdaBoostM1 |           10 |    0.0010456 |           31 |
|   16 | Accept |        0.45 |     0.20565 |        0.45 |     0.44993 |  GentleBoost |           10 |       0.9156 |           99 |
|   17 | Accept |        0.45 |      0.1802 |        0.45 |     0.44993 |   AdaBoostM1 |           10 |    0.0091856 |           98 |
|   18 | Accept |        0.56 |      0.2018 |        0.45 |     0.44996 |   AdaBoostM1 |           10 |      0.43865 |           11 |
|   19 | Best   |        0.42 |     0.20737 |        0.42 |     0.42002 |  GentleBoost |           10 |      0.98704 |           56 |
|   20 | Accept |       0.445 |     0.25983 |        0.42 |     0.43315 |  GentleBoost |           14 |      0.77577 |           50 |
|===================================================================================================================================|
| Iter | Eval   | Objective   | Objective   | BestSoFar   | BestSoFar   |       Method | NumLearningC-|    LearnRate |  MinLeafSize |
|      | result |             | runtime     | (observed)  | (estim.)    |              | ycles        |              |              |
|===================================================================================================================================|
|   21 | Accept |        0.43 |     0.22077 |        0.42 |     0.43172 |  GentleBoost |           11 |      0.73441 |           64 |
|   22 | Accept |        0.45 |     0.23336 |        0.42 |     0.43652 |  GentleBoost |           12 |      0.96312 |           65 |
|   23 | Best   |       0.395 |     0.23522 |       0.395 |     0.42961 |  GentleBoost |           12 |      0.87379 |           56 |
|   24 | Accept |        0.53 |     0.20113 |       0.395 |     0.42991 |   AdaBoostM1 |           10 |      0.83388 |            3 |
|   25 | Accept |         0.5 |     0.21561 |       0.395 |     0.41094 |   AdaBoostM1 |           10 |    0.0010431 |            1 |
|   26 | Accept |       0.455 |     0.22147 |       0.395 |     0.43429 |  GentleBoost |           11 |      0.63755 |           57 |
|   27 | Accept |       0.505 |     0.21564 |       0.395 |     0.43441 |  GentleBoost |           10 |      0.93185 |            2 |
|   28 | Accept |        0.51 |     0.21469 |       0.395 |     0.43429 |  GentleBoost |           10 |    0.0011029 |           21 |
|   29 | Accept |        0.52 |     0.21221 |       0.395 |     0.43435 |  GentleBoost |           10 |      0.94349 |            1 |
|   30 | Accept |        0.45 |     0.19013 |       0.395 |     0.43428 |   AdaBoostM1 |           10 |      0.94065 |           99 |

__________________________________________________________
Optimization completed.
MaxObjectiveEvaluations of 30 reached.
Total function evaluations: 30
Total elapsed time: 28.5134 seconds
Total objective function evaluation time: 19.155

Best observed feasible point:
      Method       NumLearningCycles    LearnRate    MinLeafSize
    ___________    _________________    _________    ___________

    GentleBoost           12             0.87379         56     

Observed objective function value = 0.395
Estimated objective function value = 0.43428
Function evaluation time = 0.23522

Best estimated feasible point (according to models):
      Method       NumLearningCycles    LearnRate    MinLeafSize
    ___________    _________________    _________    ___________

    GentleBoost           12             0.87379         56     

Estimated objective function value = 0.43428
Estimated function evaluation time = 0.23364

predictedSpecies = kfoldPredict(rfModel);

Incorrect number or types of inputs or outputs for function kfoldPredict.

confusionchart(response,predictedSpecies)

Answer 2

[Walter Roberson]

When you specify method 'Bag' for fitcensemble() then the output is a ClassificationBaggedEnsemble .

kfoldPredict() does not accept bagged ensembles. https://www.mathworks.com/help/stats/classreg.learning.partition.classificationpartitionedmodel.kfoldpredict.html#bsu1ryc-1_sep_mw_5cb12047-1d4f-4f41-8dfb-c81f28235bb5

[evelyn]

thank you for your reply!

Is there any way that I can do kfold with 'ClassificationBaggedEnsemble'?

[Walter Roberson]

I am not certain, but it appears to be that when you use Bag that the output model is not a "cross-validated partitioned classifier", so I suspect that the answer is No (at least at present.)