MackNet_Fit: MackNet_Fit

In EduardoRamosP/MackNet: Mack-Net model: Blending Mack's model with Recurrent Neural Networks

Description

This function fits the ensemble of RNNs required for the payments MackNet model. The optimum weigthed decay is obtained by selecting the configuration that minimizes the test error.

Usage

MackNet_Fit(
  Cumulative.T,
  Incurred.T,
  Exposure,
  AR,
  Ensemble,
  wd,
  Learning,
  drop,
  Epochs,
  MinimumEpochs,
  ES,
  Output
)

Arguments

Cumulative.T	Cumulative payments triangle.
Incurred.T	Incurred cost triangle.
Exposure	Exposure measure. Written premiums is an appropriate measure to scale cumulative payments and incurred cost.
AR	This variable allows to remove the autorregressive component from the MackNet model when it is set to 0.
Ensemble	Number of RNNs included in the ensemble.
wd	The optimization algorithm used is ADAM. This variable defines the weighted decay value.
Learning	Learning rate.
drop	Dropout regularization.
Epochs	Maximum number of epochs.
MinimumEpochs	Minimum number of epochs.
ES	Early Stopping object defined under the keras framework.
Output	Linear or ReLU activation function for the output layer.

Value

The formula generates the following outputs:

• TrianglesBackup Full triangles predicted by each RNN included within the ensemble.

• Error Test error.

EduardoRamosP/MackNet documentation built on Sept. 26, 2020, 9:21 a.m.