Package 'agena.ai'

Calculate posterior probabilities in a BN on agena.ai cloud

Description

A function to send an input Bayesian network model to Agena AI Cloud servers. Once called, the function will check authentication status, if it has not expired, it will send the POST request with the model to the servers, and receive calculation results to update the Bayesian network model (filling the results field with calculation results).

Usage

calculate(input_model, login, dataSet = NULL, debug = FALSE)

Arguments

input_model	an R model object
login	an agena.ai cloud login
dataSet	a dataSet in the R model object
debug	boolean parameter to display debug messages or not

Value

BN inference results in the model

---

Create a json file of a BN model with observations

Description

This function takes an R Model objectand a CSV file of observations for cases and creates a batch of datasets / cases for each row in the input data and generates a .json file for the BN

Usage

create_batch_cases(inputModel, inputData)

Arguments

inputModel	an R model object
inputData	CSV file of observations

Value

json file

---

Create a CSV template for a model for batch observations

Description

This function creates an empty CSV file with the correct format so that it can be filled in and used for create_batch_bases.

Usage

create_csv_template(inputModel)

Arguments

inputModel

an R model object

Value

csv file

---

Create configuration object for sensitivity analysis

Description

See agena.ai manual for sensitivity analyis configuration

Usage

create_sensitivity_config(
  target,
  sensitivity_nodes,
  dataset = NULL,
  network = NULL,
  report_settings = NULL
)

Arguments

target	target node in the network
sensitivity_nodes	a set of sensitivity nodes in the network
dataset	a dataSet in the model
network	a network in the model
report_settings	a list of report settings

Value

sensitivity config object

---

Dataset

Description

Dataset

Fields

id

dataSet id

observations

list ob observations.

results

list of inference results.

---

Load a Model from CMPX

Description

This function reads an input agena.ai model file with the .cmpx extension to create an R model object. The model object includes the networks and the nodes of the imported model.

Usage

from_cmpx(modelPath)

Arguments

modelPath

Path to the input file

Value

An R model object

---

Generate a CMPX file from R Model

Description

This function generates a .cmpx file for an agena.ai model based on an R model object

Usage

generate_cmpx(inputModel)

Arguments

inputModel

an R model object

Value

cmpx file

---

Local agena.ai developer license activation

Description

sends the agena.ai developer license for activation

Usage

local_api_activate_license(key)

Arguments

key	agena.ai developer license key

Value

no return value, activates user's local API license

---

Local agena.ai API model calculation for batch cases

Description

Runs inference for every dataSet in the model

Usage

local_api_batch_calculate(model)

Arguments

model

an R model object

Value

a json file for the model with results

---

Local agena.ai API model calculation

Description

Local agena.ai API model calculation

Usage

local_api_calculate(model, dataSet, output)

Arguments

model	an R model object
dataSet	a dataSet in the model
output	file name for the output json - just the filename using the working directory, not a full file path

Value

a json file for the model with results

---

Local agena.ai API clone

Description

clones the local agena.ai developer API in the working directory

Usage

local_api_clone()

Value

no return value, used to clone local API to directory

---

Local agena.ai API compilation

Description

sets the version and compiles the maven environment for the local agena.ai developer API in the working directory

Usage

local_api_compile()

Value

no return value, used to compile maven in the local API directory

---

Local agena.ai API sensitivity analysis

Description

Local agena.ai API sensitivity analysis

Usage

local_api_sensitivity(model, sens_config, output)

Arguments

model	an R model object
sens_config	a sensitivity config object
output	file name for the output json - just the filename using the working directory, not a full file path

Value

a json file report of the sensitivity analysis results

---

Login to agena.ai cloud

Description

Login to agena.ai cloud

Usage

login(username, password)

Arguments

username	agena.ai cloud username
password	agena.ai cloud password

Value

no return value, used in to authenticate user credentials

---

BN Model object

Description

BN Model object

Fields

id

model id

networks

list of networks in the model

dataSets

list of dataSets in the model

networkLinks

list of network links in the model

settings

list of settings of the model

Methods

add_network(newNetwork)

A method to add a new Network object to the networks field of a Model object. The input newNetwork is a Network object and it is added to the model if it is not already in it.

add_network_link( source_network, source_node, target_network, target_node, link_type, pass_state = NULL )

This is the method to add links to a model between its networks. These links start from a "source node" in a network and go to a "target node" in another network. To create the link, the source and target nodes in the networks need to be specified together with the network they belong to (by the Node and Network ids). See README.md for details.

change_settings(settings)

A method to change model settings. The input parameter settings must be a list with the correctly named elements, see README.md for example.

clear_all_observations()

A method to clear all observations defined in a model. This function removes all observations from all datasets (scenarios).

clear_dataSet_observations(dataSet)

A method to clear all observations in a specific dataset (scenario) in the model.

create_dataSet(id)

It is possible to add multiple scenarios to a model. These scenarios are new DataSet objects added to the dataSets field of a model. Initially these scenarios have no observations and are only defined by their ids. The scenarios are populated with the enter_observation() function.

default_settings()

A method to reset model settings back to default values.

enter_observation( dataSet = NULL, node, network, value, variable_input = FALSE, soft_evidence = FALSE )

A method to enter observation to a model. To enter the observation to a specific dataset (scenario), the dataset id must be given as the input parameter dateSet. If dataSet is left NULL, the entered observation will by default go to "Scenario 1". This means that if there is no extra datasets created for a model (which by default comes with "Scenario 1"), any observation entered will be set for this dataset. See README.md for details and examples.

get_dataSets()

A method to list the ids of all existing scenarios in a model.

get_networks()

A method to see ids of all the networks in a model.

get_results(filename = NULL)

A method to generate a .csv file based on the calculation results a Model contains.

import_results(result_file)

A method to import results of a calculated dataSet from a json file. This correct format for the results json file for this method is the file generated with the local agena.ai developer API calculation (see README.md for details.

initialize( id = NULL, networks, from_cmpx = FALSE, dataSets = NULL, networkLinks = NULL, settings = NULL )

Creates a new Model object, a list of networks (at least one) is required. id and settings are set with defaults unless specified.

remove_all_network_links()

A method to remove all existing network links in a model.

remove_dataSet(olddataSet)

A method to remove an existing scenario from the model. Input parameter olddataSet is the string which is the id of a dataset (scenario).

remove_network(oldNetwork)

A method to remove an existing Network object from the model. Note that removing a Node from a network does not automatically remove its possible network links to other networks in the model. networkLinks field of a Model should be adjusted accordingly if needed.

remove_network_link(source_network, source_node, target_network, target_node)

A method to remove network links, given the ids of the source and target nodes (and the networks they belong to).

remove_observation(dataSet = NULL, node, network)

A method to remove a specific observation from the model. It requires the id of the node which has the observation to be removed and the id of the network the node belongs to.

to_cmpx(filename = NULL)

A method to export the Model to a .cmpx file. This method passes on all the information about the model, its datasets, its networks, their nodes, and model settings to a .cmpx file in the correct format readable by agena.ai. If the input parameter filename is not specified, it will use the Model$id for the filename.

to_json(filename = NULL)

A method to export the Model to a .json file. This method passes on all the information about the model, its datasets, its networks, their nodes, and model settings to a .json file in the correct format readable by agena.ai. If the input parameter filename is not specified, it will use the Model$id for the filename.

---

BN Network object

Description

These represent each network in a BN. Networks consist of nodes and in a BN model there might be more than one network. These networks can also be linked to each other with the use of input and output nodes. For such links, see Model$networkLinks.

Fields

id

network id

name

network display name

description

network description

nodes

list of nodes in the network

Methods

add_node(newNode)

A method to add new Node objects to a Network. Note that adding a new Node to the network does not automatically add its parents to the network. You need to add all the parents separately too.

initialize(id, name = NULL, description = NULL, nodes = NULL)

Creates a new Network object, a unique id is required, other fields are filled with defauls unless specified.

plot()

A method to plot the graphical structure of a BN network.

remove_node(oldNode)

A method to remove an existing Node object from the network. Note that removing a Node from a network does not automatically remove it from its previous parent-child relationships in the network. You need to adjust such relationships separately on the Node level.

---

BN Node object

Description

These represent the nodes in a BN.

Fields

id

node id

name

node display name

description

node description

type

node type

parents

node parent nodes

simulated

whether node is simulated

distr_type

node distribution type

states

node states

probabilities

node probabilities

expressions

node expressions

partitions

node expression partitions

variables

node variables or constants

Methods

addParent_byID(newParentID, varList)

This is a method to add parent Nodes by their ids for cmpx parser capabilities. To add parents to Node objects, please use $add_parent(Node) method.

add_parent(newParent)

Adds a Node object as a new parent node to the current Node object and resets/resizes the NPT values and expressions of the Node as needed. Parents list of a Node object is a list of other Node objects. The input parameter of the function is a Node object variable. A good practice is to use Node ids as their variable names.

get_parents()

A method to list all the existing parent nodes of a Node.

initialize( id, name = NULL, description = NULL, type = NULL, simulated = FALSE, states = NULL )

Creates a new Node object, a unique id is required, other fields are filled with defaults unless specified. Node id, name, description, type, states, and whether it is a simulation or regular node do not depend on its edges and parents in a network, a Node object can be defined with all this information outside a Network as well. To add/remove/modify parents, distr_type, probabilities, expressions, and partitions; use the corresponding method.

remove_parent(oldParent)

Removes a Node object from parents of the current Node object and resets/resizes the NPT values and expressions of the Node as needed. The input parameter of the function is a Node object variable. A good practice is to use Node ids as their variable names.

remove_variable(variable_name)

A method to remove one of the existing variables (constants) from a node, using the variable_name.

set_distribution_type(new_distr_type)

A method to set the table type (distr_type) of a node. If a Node is simulated, its table type can be "Expression" or "Partitioned" - the latter is only if the node has parent nodes. If a Node is not simulated, its table type can be "Manual", "Expression", or "Partitioned Expression (if the node has parent nodes)".

set_expressions(new_expr, partition_parents = NULL)

The method to set the probability values if the table type (distr_type) of a Node is "Expression" or "Partitioned". If the table type is "Expression", new_expr is a single string and partition_parents is left NULL. If the table type is "Partitioned", new_expr is a list of expressions for each parent state, and partition_parents is a list of strings for each partitioned parent node id.

set_probabilities(new_probs, by_rows = TRUE)

The method to set the probability values if the table type (distr_type) of a Node is "Manual". new_probs is a list of numerical values, and the length of the input list depends on the number of the states of the node and of its parents.

You can format the input list in two different orders. If the parameter by_rows is set to true, the method will read the input list to fill in the NPT row by row; if set to false, the method will read the input list to fill in the NPT column by columnn. See README.md for examples.

set_variable(variable_name, variable_value)

A method to set variables (constants) for a node. Takes the variable_name and variable_value inputs which define a new variable (constant) for the node.

---

Sensitivity analysis on agena.ai cloud

Description

Sensitivity analysis on agena.ai cloud

Usage

sensitivity_analysis(input_model, login, sensitivity_config, debug = FALSE)

Arguments

input_model	an R model object
login	an agena.ai cloud login
sensitivity_config	a sensitivity analysis config object
debug	boolean parameter to display debug messages or not

Value

sensitivity analysis report

---