0
$\begingroup$

How the neural networks are overfitted for regression.Either it tries

to equal individual observation values or equals to the sum of all observations

$\endgroup$

1 Answer 1

2
$\begingroup$

Neural Networks basically act as a high memory-based machine learning algorithm. So for a given dataset the chance of it perfectly aligned with all the data at a given instance is high, as it most likely just ends up remembering every data point you give.

Overfitting occurs precisely because of this, when a new expansive data set is introduced there is no way it can adjust its fit to the new data, graphically it ends up missing more of the values than its supposed to fit.

In conclusion, it does not work well in the case where the scoring population is significantly different compared to training sample.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.