Question

State in your own words what supervised and unsupervised learning is.

Clearly describe a real-world scenario where each classifier would be useful.

Answer 1

We come across the words supervised learning and unsupervised learning in data mining.

Supervised learning :

      Supervised learning simply means getting correct output from the given data based on already predefined data (trained data).

•       If we give new data as a input to the algorithm of supervised learning , then the algorithm     analyzes the trained data and by comparing with the trained data , new data is processed and classified.
•       Trained data consists of many examples with predefined statements and no two examples consists all the statements or features as same.
•       For instance we have different types of fruits with us.Each fruit is taken as one example and we define physical features of that fruit in that example. For instance apple has physical features like

             color : red

                   Shape :sphere with depression at the top.

           Banana has physical features like

             color : yellow

             shape : cylindrical shape with some curve.

•      All these examples of fruits present in our trained data.
•      Now if user gives a fruit as a input and we check physical features of the given fruit with the trained data and we give the name of the given fruit as output.
•      Few more examples are identifying the disease of a person based on the symptoms given in trained data.

Unsupervised learning :

     Unlike supervised learning we don't have any trained data and we should design our algorithm of unsupervised learning in such a way that it itself classify or group the given data based on the similarities.

•      For instance we have given some fruits and we don't have any trained data to classify that fruits.
•      So suppose we first group the fruits based on the shape.
•      we have pomegranate and guava under same group because both have spherical shape.
•      Next we differentiate based on color i.e., pomegranate is red in color and guava is green in color.
•      In this way we have unsupervised learning.
•      Clustering is a problem which comes under unsupervised learning . One example of clustering problem is grouping the customers based on their behavior of purchasing.

Some differences between supervised and unsupervised learning :

•       Since we are having trained data in supervised learning we initially knows number of classes or groups . But in unsupervised learning we can know no of groups or classes only after processing the input data.
•       Supervised learning results are accurate and reliable because results are completely based on trained data. Whereas in unsupervised learning we are not sure that we get completely accurate and reliable results.