I tried to implement Polynomial Regression without scikit . But Whenever i try to minimize my cost function using gradient descent , it decreases but it never saturates no matter how many iterations i use gradient descent on . Can someone suggest what could be the possible reasons for it ?
Tips:
- Normalize the data set before training.
- Try different learning rates.
- Try using more or fewer iterations.
If you want more suggestions, please post a plot of the cost history during training.
Also, when you say “not saturated”, I think you meant “did not converge to a stable minimum value”?
@TMosh , yes that is what i mean when i say not saturated .
and this is my cost function , please take a look at it .
The cost is decreasing, so that’s good.
Have you tried the issues I suggested?