RapidMiner

manual calculation of NeuralNet predictions

SOLVED
Contributor II

manual calculation of NeuralNet predictions

Hi,

Can someone pls let me know how to calculate a prediction (continuous) from a set of input variables (continuous) using the weights and bias at hidden nodes (sigmoid) and the output node (linear)? This is to manually verify the results obtained from neuralnet.

Thanks

 

1 ACCEPTED SOLUTION

Accepted Solutions
Highlighted
Contributor II
Solution
Accepted by topic author darshi_plm
‎01-16-2017 08:08 PM

Re: manual calculation of NeuralNet predictions

Thanks a lot.
I have used the NeuralNet output weights to calculate the sigmoid transformation of weighted sum of the inputs. The mistake I have made was calculating the output without realizing that NeuralNet has normalised the attributes to fall within -1 and 1. Now all sorted.

2 REPLIES
Moderator

Re: manual calculation of NeuralNet predictions

Highlighted
Contributor II
Solution
Accepted by topic author darshi_plm
‎01-16-2017 08:08 PM

Re: manual calculation of NeuralNet predictions

Thanks a lot.
I have used the NeuralNet output weights to calculate the sigmoid transformation of weighted sum of the inputs. The mistake I have made was calculating the output without realizing that NeuralNet has normalised the attributes to fall within -1 and 1. Now all sorted.