ML | K-NN

BASICS
KNN is a classification algo for SuperVised Learning
K - nearest neighbour

if we have 2 type of class and a point out of class
we have To tell which class that [oint belongs then we use K-NN algo

K is the parameter which tell how many near neighbour we have to see
if the max near neighbour is of class A then the point will belong to
class A

 generelly used for small dataset
 K = squt of all the numbers of point (all take odd value of K)

 we plot he Validation and error curve of our data Set :
 the minimum value of K for both the curve is taken as value of K

PSEUDO CODE
1 Load the data
2 Initialise the value of k
3 For getting the predicted class, iterate from 1 to total number of training data points
    Calculate the distance between test data and each row of training data.
    Here we will use Euclidean distance as our distance metric since it’s the most
    popular method. The other metrics that can be used are Chebyshev, cosine, etc.
    Sort the calculated distances in ascending order based on distance values
    Get top k rows from the sorted array
    Get the most frequent class of these rows
    Return the predicted class of max frequent

The point whose class is to be determined : we calculate its distance from all the point of classes
sort the distance in ascending order
and now we take K numbers from this sorted vector  (neighbour)
and check the class of each member in neighbour
Get the most frequent class in this neighbour and this wil be predicted class

WHAT WE HAVE TO DO?

choose the numbers of K neighbour
take the k nearest neighbour of the new data point according to the Euclidean distance
Among the K neighbour count the numbers of data point in each category
Assign the new data point to the category where you counted the most neighbour