python code for Keras Deep NN Confidence Interval for regression using
$10-30 USD
Completed
Posted almost 3 years ago
$10-30 USD
Paid on delivery
data is tabular with both categorical and continues features
target is continues so it is regression
you need to simulate data or find existing data set
data you create by yourself or you existing data set
but both categorical and continues features
data is not big - lets say 100000 rows and 18 features
8 features categorical and 8 features continues
many values for each categorical feature
you need some special way to find reliability for each predictions
but use many time runs with different random seed - use something better and smarter
for example do not use this approach
[login to view URL]
[login to view URL]
since approach is weak one
you need to find better solution
maybe
[login to view URL]
[login to view URL]
[login to view URL]@qucit/a-simple-technique-to-estimate-prediction-intervals-for-any-regression-model-2dd73f630bcb
[login to view URL]
[login to view URL]:1472065/[login to view URL]
page 14
2.4.4 Prediction Interval for Artificial Neural Networks Several methods to estimate prediction intervals for neural networks exist. These include dropout [4], bootstrap [23], mean variance estimation [23] and upper lower bound estimation [24].
[login to view URL]
[login to view URL]@qucit/a-simple-technique-to-estimate-prediction-intervals-for-any-regression-model-2dd73f630bcb
algorithm from these links maybe not good enough, you need to find theory and develop code to provide good solution