For example, it is easy to predict a numeric out put from with a categorical data (numerical patterns). But if the out put that needs to be predicted is a pattern (numerical vector), which is the best way in machine learning?
Hello Adithya Challapalli ,
there are several methods to do that. I would start with an artificial neural network, in which the output layer has a number of nodes equal to the length of the pattern that you want to predict.
Regards
yes, we predict a numeric pattern from numeric input using machine learning algorithm and then there are several algorithm we are using to a numerical pattern
refer this link how to use to get some ideas
https://www.oreilly.com/ideas/uncovering-hidden-patterns-through-machine-learning
Jin Li Félix Leaman Karthika .P
Hello all,
Thanks for the suggestions. I've tried random forest and ANN using matlab, but they say that a floating numerical output variable is required. I could not figure out how to use this models to predict categorical variables. Can you please provide me with any examples please. My problem is a regression type, with supervised learning and most importantly the output(y) is a categorical variable (a numerical combination of vectors)
Why don't you change the categorical variable into a number. For example, if you have gender as an attribute, then turn it into two attributes (male, and female). Now, if the record is gender = male, it becomes male = 1 and female = 0.
Now you can use RF or NN to fit.
Mohammed Ghazal
Yes, I actually did. I got around 14 different categories (numbers). But my problem is to do regression. It works fine if I use the categorical variable which I encoded as numerical data as input. But I am having trouble training the categorical variables as output.
If the output (target, class,) variable is categorical, then it is a classification not a regression problem. No?
Mohammed Ghazal
Kind off, but take this as a combination problem. I need to find the best combination as output, but depending on numerical data
- Badges
- Science topic