Hi every one.,
I'm working on multi-class data, and I want to predict the test data. In this regard, I have problem with introduction the response matrix to the classifer. The most common methods are adapted with two classes (0, 1) but about multi-class data there are lots of limitations...
Can any one help me?
Many Thanks
Reihaneh
There are very simple classifiers that solve multi-class problems. For instance, you may first try the k-NN method. There are implementations of k-NN in any programming language. Other multi-class classifiers that are very simple are Naive Bayes and C4.5 (decision tree).
Please refer to the following link
http://www.mit.edu/~9.520/spring09/Classes/multiclass.pdf
Hi, That is an easy task. By PLSDA you can use multicalss just as you apply PLS2, you arrange your class label in this format,:::::::
Object 1 for class 1 1 0 0 0
Object 2 for class 2 0 1 0 0
.
.
.
.
.
Object n for class 3 0 0 1 0
and so on.....................
Dear Dr Vasighi, Eduardo, Mahboobeh, Harsh and Prof Ghasemi
Many Many Thanks for your kind response.
Kindest regards
Well day
Reihaneh
Hi, besides other mentioned methods above, in my point of view, soft independent modelling by class analogy (SIMCA), as one of the most common and simplest classifiers in pattern recognition context, would be so advantageous in challenging multi-class states during classification modeling.
This method performs based on PCA method so that for each class we define an individual PCA model with a specific number of principal components describing the data space (distribution of objects) in the target class. It might be helpful when we encounter with overlapping between classes in data space.
Best wishes
Neural Network follows a softmax classifier may solve your problems. You can refer this,http://deeplearning.stanford.edu/wiki/index.php/Softmax_Regression.
Hi Mr. Shahbazy,
As previous researchers said, it is a rather easy task. First of all you can use Linear Discriminant Analysis (LDA) approach and only define your class vector in following order:
Y= 1;1;1;1;1;1;1;1;2;2;2;2;2;2;2;3;3;3;3;3;3;3;4;4;4;4;4;4;4, respect to the row order of the objects in X space,
Then, using the following command it would be possible to run multiclass classification approach in MATLAB.
[Yhat,e]=classify(X,X,Y);
Yhat is the predicted class vector and e is the classification error.
For those classes that are not seperated linearly it would be possible to implement quadratic discriminant analysis (QDA) using the following command:
[Yhat,e]=classify(X,X,Y,'quadratic');
Finally, as Dr. Vasighi said, using one against all and one against one approaches and the majority voting strategy works better than the multi-class classification procedure.
Please see:
http://www.sciencedirect.com/science/article/pii/S0031320306002081
Good Luck ;)
Dear Mohammad, Boyu and Piyabute,
Thank you so much.
All the Best
Reihaneh
I suggest Random-Forests which is a ensemble extension for Decision trees, it's well suited for MultiClass classification problems.
Here's an introductory tutorial: http://research.microsoft.com/apps/pubs/default.aspx?id=155552
Matlab toolbox for Random-Forests: https://github.com/jrderuiter/randomforest-matlab
- Badges
- Science topic