And beware over-fitting your model! unless the problem has an absolute equation, if the answer is almost perfect (i.e the errors are very small), then its over-fitted and the model is useless. Its useless, because in the real world nothing is absolute, even the measurements of radius & circumference of a circle contain errors so all the data points have there own little measurement errors. Well trained AI / Machine learning models use hold out data samples from which the model doesn't learn, so it can 'test' the model after the learning is complete. Reading Ian's The Average is Always Wrong is a great way to get into and demystifies the topic.
And beware over-fitting your model! unless the problem has an absolute equation, if the answer is almost perfect (i.e the errors are very small), then its over-fitted and the model is useless. Its useless, because in the real world nothing is absolute, even the measurements of radius & circumference of a circle contain errors so all the data points have there own little measurement errors. Well trained AI / Machine learning models use hold out data samples from which the model doesn't learn, so it can 'test' the model after the learning is complete. Reading Ian's The Average is Always Wrong is a great way to get into and demystifies the topic.