Dataset –
The dataset used for this project is the LETOR (Learning to Rank using Regression) provided by Microsoft. This dataset consists of 46 features which are ranked on a scale of 0 to 2. The scale is the measure of the rank given to a particular website. For the project we process the dataset into target vectors and training vectors. The target vectors are the ranking labels and the training vectors are the 46 features with 69,622 samples of all 46 features.
Approach –
The problem is solved using two different approaches. The two approaches are the closed form solution and Stochastic Gradient Descent (SGD). The closed form solution is a defined mathematically equations and are finite. The SGD is a gradient descent method to calculate the updated weights by differentiation.