KNN

Question 1

full KNN form

Answer 1

K nearest neighbour

Question 2

k value

Answer 2

odd/even

Question 3

how to decide k value

Answer 3

‘k’ value will be decided based on the no. of classes present in the output parameter gf training data set

Question 4

When will be k even and when it will be odd

Answer 4

If training data set contain even no. of class labels then ‘k’ value is an odd no.
If training data set contain odd no. of class labels then ‘k’ value is an even no.

Question 5

k formula

Answer 5

k = n+1 [n: number of class labels]

Question 6

knn algo (3 other properties of algo i.e. type of algo)

Answer 6

KNN algo is:
* Non parametric learning algo
* Instance based learning algo
* Lazy learning algo

Question 7

Non parametric algo
also example

Answer 7

Knn is a non parametric algo because this model is never trained with an input parameter
Non parametric algo in ML are methods that do not assume a fixed form or finite set of parameters for the model structure. Instead they allow to the model to grow in complexity with the size of the training data.
Ex: KNN, Decision trees, SVM

Question 8

Instance based algo

Answer 8

KNN is an instance based learning algo means given new observation (test data/test instance) comapred with all the training data set samples to make the predictions.
Instance based learning also known as lazy learning algorithms are class of ML techniques that delay the process of generalizing the training data until a query is made. Unlike eager learning algorithms which build a model during the training phase, instance based learning store the instances and use them directly to make predictions.

Question 9

Lazy learning algo

Answer 9

KNN algo is lazy learning algo means no model is trained to make the predictions.
Instance based learning also known as lazy learning algorithms are class of ML techniques that delay the process of generalizing the training data until a query is made. Unlike eager learning algorithms which build a model during the training phase, instance based learning store the instances and use them directly to make predictions.

Question 10

KNN - Classification / Regression

Answer 10

Both classification and regression

Question 11

note about training data set in knn

Answer 11

This algo is continuously updated with a new training data set

Question 12

When is knn used?

Answer 12

KNN is used when the relationship between feature is complex not linear

Question 13

Hamming Distance

Answer 13

If x1 and x2 are the attribute value of two instances then if x1==x2 then hamming distance is 0 else if x1≠ x2 then distance is 1

hamming distance is case sensitive

Question 14

KNN output diagram

Answer 14

Question 15

Que1 of KNN

Answer 15

class I