Welcoem to posts!!

Learn with Vichu
Posted 1 year ago

How Linear Regression Works?

The linear regression algorithm tries to fit a straight line to the data points which approximate the data in a best way.

For that it uses the straight line formula which is Y = MX + C.

During the training process the algorithm tries to learn the optimal values for weight and bias, for that it uses the concept called gradient decent.

The idea behind gradient decent is , in the beginning time weight and bias will be initialised randomly or it could be zeros.

After initialising weights and bias the target values will be calculated using the straight line formula and loss will be computed using loss function like mean squared error, root mean squared error, or mean absolute error.

After calculating the loss the algorithm will try to update the weights and bias using gradient. To get gradients it will differentiate loss function with respect to weight and bias.

Then the calculated gradients will be used to update weights and bias.

This process will be continued for n iterations.

2 - 0