Boston house price prediction

#1

i am getting 0% accuracy on NN can you tell why and how can i correct it?
link:https://drive.google.com/open?id=1PeMjhpS7Pmen0LNESqu_MwpAgR_xZa6t

#2

Hey @footballer.salik, please upload your ipynb notebook only (not in doc format) so that it contains outputs as well and than share the link.

#3

ipynb link:https://drive.google.com/file/d/1RysQs7yUdwEJaHu_JDtroGEgV1qRDtqc/view?usp=sharing

#4

Hey @footballer.salik, you have used wrong loss function, since this is a regression problem you must use ‘mean_squared_error’. Also for calculating accuracy you can use ‘mae’ which stands for mean absolute error.

Hope this cleared your doubt. :blush:

#5

stil i am getting negative accuracy for test data .can you tell what i have done wrong?
link:https://drive.google.com/file/d/1RysQs7yUdwEJaHu_JDtroGEgV1qRDtqc/view?usp=sharing

#6

That is because you are again using
model.compile(optimizer=‘Adam’,loss = ‘binary_crossentropy’,metrics=[‘accuracy’])

which should be
model.compile(optimizer=‘Adam’,loss = ‘mean_squared_error’,metrics=[‘mae’])

#7

sorry i have sent the wrong link
new link:https://drive.google.com/file/d/1FxVm7-6InqehlMDq1do6Yu7iX45tGyCR/view?usp=sharing

#8

Also now validation results are going in right direction.Now your model is quite basic, increase the number of layers and try out various things, change activation function, that is why the problem is a part of leaderboard challenge. Its not ethical to directly know the answer from TA’s. We are here to guide you but you need to walk yourself.

Happy Learning :sunny: