compute.ann: Neural Network-based Interaction Index for a Quarry

In quarrint: Interaction Prediction Between Groundwater and Quarry Extension Using Discrete Choice Models and Artificial Neural Networks

Description

Given an object of type quarry, a neural network computes the interaction index (low, medium, high or very high).

Usage

## S3 method for class 'quarry'
compute.ann(x, ann = NULL, rep = 1, ...)

Arguments

x	A quarry object.
ann	The neural network used to estimate the interaction index. By default, if set to NULL, it uses a neural network provided by the package. The neural network is an object of class nn.
rep	The repetition of ann to be used.
...	Further arguments passed to or from other methods. For instance, see compute of the package neuralnet.

Details

The neural network provided with the package has a feed-forward design and has a hidden layer of 7 nodes. It takes as an input a quarry constructed with the function "quarry" using the default parameters. This neural network is fully detailed in "Interaction prediction between groundwater and quarry extension using discrete choice models and artificial neural networks" (Barthelemy et al., 2016).

It is possible to use another neural network that has been trained with the function train.ann.

Value

A list whose elements are:

low	The output of the ann for a low interaction level.
medium	The output of the ann for a medium interaction level.
high	The output of the ann for a high interaction level.
very.high	The output of the ann for a very high interaciton level.
idx	a string with the level of interaction (low, medium, high or very high).

Note

The quarry x must have been created by the quarry. It can accept custom ranges for the parameters values but they must be consistent with what has been used to train the neural network ann.

Author(s)

Johan Barthelemy.

Maintainer: Johan Barthelemy [e-mail].

References

Barthelemy, J., Carletti, T., Collier L., Hallet, H., Moriame, M., Sartenaer, A. (2016) Interaction prediction between groundwater and quarry extension using discrete choice models and artificial neural networks Environmental Earth Sciences (in press)

Collier, L., Barthelemy, J., Carletti, T., Moriame, M., Sartenaer, A., Hallet, H. (2015) Calculation of an Interaction Index between the Extractive Activity and Groundwater Resources Energy Procedia 76, 412-420

Krieselm, D. (2007) A Brief Introduction to Neural Networks. On-line available at http://www.dkriesel.com

Ripley, B. (1996) Pattern recognition and neural networks Cambridge university press

See Also

compute.dc to compute an interaction index based on a discrete choice model and compute.interaction to predict the interaction between between the quarry and the groundwater using both the discrete choice-based index and the neural network-based index.

train.ann to train a neural network and use it as an input for this function.

compute and neuralnet of the package neuralnet for optional additional parameters and details about objects of class nn.

Examples

# creating a quarry
q <- quarry(geological.context = 2, hydrogeological.context = 4,
            piezometric.context = 1, quarry.position = 4,
            production.catchment = 4, quality.catchment = 3)

# computing the interaction index using the default neural network
inter.idx <- compute.ann(q)

quarrint documentation built on May 1, 2019, 10:10 p.m.