Supervised Learning - Prediction Week 3 Challenge

Q2.Difference between L1 and L2 gradient descent method.

Answer:-I) 1 :- In absolute mean Error function is give the proper output

2 :-L1 is Absolute mean square function is more robust and is generally not affected by outliers.
3 :-Adds penalty in terms of cost function this term is absolute sum of the coefficient.
4 :-it tends to make coefficient to absolute zero.

II) 1 :-L2 is a mean square Error function generally affected by the outliers.
2 :-Add a penalty term which is equal to sqare of the coefficient L2 term i equal to the aqure of the magnitude of the coefficient.
3 :-its never sets the value of coefficient to absolute zero.

Q3. What are the different loss function for regression.

Ans:- There are three type of loss function

1) RMSE:- Root Mean Square Function

2) MSE:-Mean Square Function

3)AMSE:-Absolute Mean Square function

Q4.What are the imporataance of learning rate.

Answer- if the learning rate is too small, learning will take so long converge. other hand , if the learning rate is too large,
during learning, it may miss the converge. it is used to scale the magnitude of parameter updates during gradient descent.
learning rate is scaler, a value thats tells the machine fast or slow to arrive at some conclusion.

Q5. How to evaluate linear regression

Answer- Using below methods we can evaluate linear regression.
Y=M*X + C
X Y XY X**2
1 3 3 1
2 4 8 4
3 5 15 9
4 7 28 16
SUM 10 19 51 30

C = (∑Y)(∑X*2) -(∑X)(∑XY)/N(∑X2)-(∑X)*2 C = 570-540/120-100 =30/20 =1.5

M = N(∑XY)-(∑X)(∑Y)/N(∑X*2)-(∑X)*2 M = 216-190/120-100 = 26/20 =1.3

Y= 1.3X+1.5

Q6.What is difference between multiple and adjusted coefficient determination.

Answer-
I)Adjusted coefficient determination:-Adjusted coefficient of determination is the adjusted value of thee coefficient of determination in which the number of variables
of the data set taken into consideration.it determines the fitting of the multiple regression equations for the sample data.

II)Multiple coefficient determination:-the term multiple coefficient of determination indicates that we are measuring the goodness of fit for the estmated multiple regression equation