Which algorithms are good for Multi class classification problems?

machine_learning
multi-class

#1

When we have more than 20 classes which algorithms are best and what is the procedure???
Please help thanks in advance…

–swapna


#2

Hi @swapna26,

Please refer this discussion thread.


#3

Hi

I want to understand when no of classes are more not features…Kindly help!!!


#4

Hey man you can use Deep Learning with Tensorflow:
Multi-class classification simple means u have more than one categorical variable and so you will use categorical_crossentropy loss fucntion and use relu and your activation function.

See an example on my github.com https://github.com/constantinembufung/multiclass-classification-with-tensorflow/blob/master/Classifying_newswires_with_Deep_Learning.ipynb


#5

Hi @constantinembufung

thanks for your reply

I am new to Data science.

Can you tell me when every row has one category for example mac_address(b4:5d:50:0e:8d:20)
i.e so many categories then how to put this field into my training data…please help


#6

You could make use of pandas data frame to do it. And add the rows using .iloc method. Please refer to the official docs here - https://pandas.pydata.org/pandas-docs/stable/generated/pandas.DataFrame.append.html


#7

@swapna26
Hi, the kind of model you want to use largely depends upon the size of the data you have if your data is huge and there are lot of features then deep neural nets are preferable but when your data is small to medium with considerable amount of features you can use any boosting algorithms or ensmeble methods.
Its not a ground rule to use deep learning in everything.