V
V
vaniavaniatf12018-11-09 17:06:43
Python
vaniavaniatf1, 2018-11-09 17:06:43

What type of neural networks should be used to predict car prices?

There is a date set with car prices. How can you predict the price?
I have an idea that you can make a regular deep network and put relu instead of softmax at the end.
Is this how it should be done, or are there other smarter approaches? What is the best way to choose the number of layers? And how much neural is needed on each of the layers?

Answer the question

In order to leave comments, you need to log in

1 answer(s)
⚡ Kotobotov ⚡, 2018-11-09
@vaniavaniatf1

Well, firstly, from the change of the separating (decisive) function, the result will not change much.
predicting prices is not the best task for neural networks - here regression models usually work, in which you can express prices in direct proportion to a set of parameters multiplied by the selected coefficients.
Neural networks usually perform better in signal classification tasks.
the number of layers is usually empirically selected

Didn't find what you were looking for?

Ask your question

Ask a Question

731 491 924 answers to any question