Regression and Classification

Regression and Classifications are two major area in Classification Technique in Data Mining.

Toady I heared a question what is regression and what is classification and where am i use which condition.

The Image says lots of word than I write.









The key point is if you have large data and you want to :

1. Find the Catagories of that You shoud follow the path via classification. 
2. You want to know the quantity you have to follow the path from regression.

Regression  : Is the task when the dicrete output y consist of one or more continuos variables.

Classification : Is the task when the dicrete output y consist of one or more dicrete varaibles.

Let's consider a input vector X1,X2,X3,........ Xn and output vector varaible Y respresenting the class varaibles Y1,Y2,Y3,.....Yn, Where n is the number of training sample.

We can express these as in mathematical function below

f(x) = y(x,w) ,where w is some adjustable parameters.

A classifier can view as the overall system that maps from X to Y.

Training is the process of finding the values for adjustable parameter w of the mapping function f on basis of the dataset. That dataset is refered as training set.

Testing preformance of a classifier is known as generalisation.

If training performance > testing performance is called overfit.

One way to reduce the overfit is divide the training data into training & validation subsets.

A well known training testing protocol for this is n-fold cross validation.