Assignment 1 Cross Validation 


Rate this product

CS5007 Deep Learning Assignment 1
Cross Validation
1 Create Dataset
1. Using the below code, create Dataset-1.
1 import numpy as np
2 import random
3 # Define input array with angles from 0 deg to 360 deg converted to radians
4 x = np . array ([ i* np . pi /180 for i in range (0 ,360 ,4) ])
5 # Setting seed for reproducibility
6 np . random . seed (10)
7 # Adding random noise to sine wave
8 y = np . sin (x) + np . random . normal (0 ,0.15 , len (x))
2. Plot Dataset-1 to visualize the points scattered roughly as a sine wave.
3. Create Dataset-2 as follows:
(a) Sample x from Gaussian Mixture Model(GMM).
(b) Sample y from N(WT X, σ)(Normal/Gaussian distribution with mean (WT X) and variance σ).
(c) Plot for different values of K(x coming form K different Gaussians).
2 Implement Ridge Regression
1. For Dataset-1,
(a) Find 15 powers of input array x and treat the whole as new input.
(b) Fit the new input and output y using ridge regression with α = 0.001.
(c) Plot the graph showing input values and fitted curve.
2. Implement Ridge Regression for Dataset-2 and plot graph.
3 Implement Cross Validation
Perform the following tasks for both Dataset-1 and Dataset-2.
1. Find the optimum value of α using k-fold cross validation manually.
2. Find the optimum value of α using k-fold cross validation using RidgeCV (Ridge regression with built-in
cross-validation) which is an inbuilt function in sklearn.
3. Plot the graph showing input values and the curve with optimum alpha found using both methods.


There are no reviews yet.

Be the first to review “Assignment 1 Cross Validation ”

Your email address will not be published. Required fields are marked *

Scroll to Top