utility-package: Construct, Evaluate and Plot Value and Utility Functions
In utility: Construct, Evaluate and Plot Value and Utility Functions
utility-package R Documentation
Construct, Evaluate and Plot Value and Utility Functions
Description
Construct and plot objective hierarchies and associated value and utility functions. Evaluate the values and utilities and visualize the results as colored objective hierarchies or tables. Visualize uncertainty by plotting median and quantile intervals within the nodes of the objective hierarchy. Get numerical results of the evaluations in standard R data types for further processing.
Details
Package: utility
Type: Package
Version: 1.4.6
Date: 2023-08-27
License: GPL-3
An objective hierarchy and an associated value or utility function
is constructed by constructing the nodes of the hierarchy starting
from the end nodes and proceeding to the higher hierarchies.
Five types of end nodes are distinguished:
End nodes of the class utility.endnode.discrete define a value
or utility function for an attribute that has a finite number of
discrete numeric or non-numeric levels.
End nodes of the classes utility.endnode.intpol1d and
utility.endnode.parfun1d implement single-attribute
value or utility functions that accept a continuous argument.
The first of these functions allows the user to specify
attribute-value pairs and performs linear interpolation between these
points.
The second function allows the user to specify any parameteric
function that is implemented as a function in R.
End nodes of the class utility.endnode.intpol2d implement
interpolated value or utility functions that are based on two
attributes.
End nodes of the class utility.endnode.cond implement
value or utility functions that assign different value or utility
functions to a finite set of attribute combinations.
End nodes of the class utility.endnode.firstavail implement
value or utility functions that try to evaluate a list of nodes and
return the value of the first node that could be evaluated based
on the provided attribute data.
Finally, end nodes of the class utility.endnode.classcounts implement
value or utility functions that value counts e.g. of species of different classes
by assigning a basic value for the occurrence of at least one species of the best class
and incrementing this value by multiplicities of species of this class and of the next lower class.
These end nodes can be implemented by using the following constructors.
utility.endnode.discrete.create
utility.endnode.intpol1d.create
utility.endnode.parfun1d.create
utility.endnode.intpol2d.create
utility.endnode.cond.create
utility.endnode.firstavail.create
utility.endnode.classcounts.create
To advance to higher hierarchical levels, values or utilities at lower
levels must be aggregated to the next higher level.
This is done ab aggregation nodes of the class utility.aggregation.
Such nodes can be implemented by using the following constructor:
utility.aggregation.create
Finally, to provide decision support under uncertainty, values at
an adequate level of the objectives hierarchy must be converted to
utilities by accounting for the risk attitude of the decision maker.
Similar to the single-attribute value or utility functions,
this can either be done by linear interpolation with a node of the class
utility.conversion.intpol or by using a parametric funciton in
a node of the class utility.conversion.parfun.
These conversion nodes can be implemented by the constructors:
utility.conversion.intpol.create
utility.conversion.parfun.create
The definition of the objective hierarchy and the associated value
and utility function can then be listed or visualized by using the
generic functions
print
summary
plot
which automaticall call the implementation corresponding to the node specified
as the first argument:
print.utility.endnode.discrete
print.utility.endnode.intpol1d
print.utility.endnode.parfun1d
print.utility.endnode.intpol2d
print.utility.endnode.cond
print.utility.endnode.firstavail
print.utility.endnode.classcounts
print.utility.aggregation
print.utility.conversion.intpol
print.utility.conversion.parfun
summary.utility.endnode.discrete
summary.utility.endnode.intpol1d
summary.utility.endnode.parfun1d
summary.utility.endnode.intpol2d
summary.utility.endnode.cond
summary.utility.endnode.firstavail
summary.utility.endnode.classcounts
summary.utility.aggregation
summary.utility.conversion.intpol
summary.utility.conversion.parfun
plot.utility.endnode.discrete
plot.utility.endnode.intpol1d
plot.utility.endnode.parfun1d
plot.utility.endnode.intpol2d
plot.utility.endnode.cond
plot.utility.endnode.firstavail
plot.utility.endnode.classcounts
plot.utility.aggregation
plot.utility.conversion.intpol
plot.utility.conversion.parfun
The value or utility function can then be evaluated by applying the
generic function
evaluate
that again calls automatically the corresponding class-specific function
evaluate.utility.endnode.discrete
evaluate.utility.endnode.intpol1d
evaluate.utility.endnode.parfun1d
evaluate.utility.endnode.intpol2d
evaluate.utility.endnode.cond
evaluate.utility.endnode.firstavail
evaluate.utility.endnode.classcounts
evaluate.utility.aggregation
evaluate.utility.conversion.intpol
evaluate.utility.conversion.parfun
This function requires the provision of observed or predicted attributes
of the valued system and returns the corresponding values or utilities
of all nodes of the hierarchy.
These results can then be visualized by providing them to the generic
function
plot
in addition to the definition of the objective hierarchy stored in the
variable corresponding to the highest node of the hierarchy.
Again, this function automatically calls the correct class-specific
implementation (the root of the hierarchy will be an aggregation
or a conversion node, not an end node):
plot.utility.aggregation
plot.utility.conversion.intpol
plot.utility.conversion.parfun
This proceedure guarantees easy handling with the simple commands
print, summary, evaluate, and plot and
the specific function descriptions provided above are only required
to check advanced attributes.
Author(s)
Peter Reichert <peter.reichert@emeriti.eawag.ch>
with contributions by Nele Schuwirth <nele.schuwirth@eawag.ch>
Maintainer: Peter Reichert <peter.reichert@emeriti.eawag.ch>
References
Short description of the package:
Reichert, P., Schuwirth, N. and Langhans, S.,
Constructing, evaluating and visualizing value and utility functions for decision support, Environmental Modelling & Software 46, 283-291, 2013.
Textbooks on the use of utility and value functions in decision analysis:
Keeney, R. L. and Raiffa, H. Decisions with Multiple Objectives - Preferences and Value Tradeoffs. John Wiley & Sons, 1976.
Eisenfuehr, F., Weber, M. and Langer, T., Rational Decision Making, Springer, Berlin, 2010.
Examples
# define discrete end node for width variability
# (attribute "widthvariability_class" with levels "high",
# "moderate" and "none")
widthvar <-
utility.endnode.discrete.create(
name.node = "width variability",
attrib.levels = data.frame(widthvariability_class=
c("high","moderate","none")),
u = c(1,0.4125,0),
names.u = c("u.high","u_moderate","u.none"),
required = FALSE,
utility = FALSE)
# define 1d interpolation end node for bed modification with
# riprap
# (attribute "bedmodfract_percent" with levels from 0 to 100)
bedmod_riprap <-
utility.endnode.intpol1d.create(
name.node = "bed modification riprap",
name.attrib = "bedmodfract_percent",
range = c(0,100),
x = c(0,10,30,100),
u = c(1,0.775,0.5625,0.24),
required = FALSE,
utility = FALSE)
# define 1d interpolation end node for bed modification with
# other material
# (attribute "bedmodfract_percent" with levels from 0 to 100)
bedmod_other <-
utility.endnode.intpol1d.create(
name.node = "bed modification other",
name.attrib = "bedmodfract_percent",
range = c(0,100),
x = c(0,10,30,100),
u = c(1,0.775,0.5625,0),
required = FALSE,
utility = FALSE)
# define combination end node for bed modification
# (attributes "bedmodtype_class" and "bedmodfract_percent")
bedmod <-
utility.endnode.cond.create(
name.node = "bed modification",
attrib.levels = data.frame(bedmodtype_class=
c("riprap","other")),
nodes = list(bedmod_riprap,bedmod_other),
required = FALSE,
utility = FALSE)
# define 1d interpolation end node for bank modification with
# permeable material
# (attribute "bankmodfract_percent" with levels from 0 to 100)
bankmod_perm <-
utility.endnode.intpol1d.create(
name.node = "bank modification perm",
name.attrib = "bankmodfract_percent",
range = c(0,100),
x = c(0,10,30,60,100),
u = c(1,0.8667,0.675,0.4125,0.24),
required = FALSE,
utility = FALSE)
# define 1d interpolation end node for bank modification with
# impermeable material
# (attribute "bankmodfract_percent" with levels from 0 to 100)
bankmod_imperm <-
utility.endnode.intpol1d.create(
name.node = "bank modification imperm",
name.attrib = "bankmodfract_percent",
range = c(0,100),
x = c(0,10,30,60,100),
u = c(1,0.775,0.5625,0.24,0),
required = FALSE,
utility = FALSE)
# define combination end node for bank modification
# (attributes "bankmodtype_class" and "bankmodfract_percent")
bankmod <-
utility.endnode.cond.create(
name.node = "bank modification",
attrib.levels = data.frame(bankmodtype_class=
c("perm","imperm")),
nodes = list(bankmod_perm,bankmod_imperm),
required = FALSE,
utility = FALSE)
# define 2d interpolation end node for riparian zone width
# (attributes "riparianzonewidth_m" and "riparianzonewidth_m")
riparzone_width <-
utility.endnode.intpol2d.create(
name.node = "riparian zone width",
name.attrib = c("riverbedwidth_m","riparianzonewidth_m"),
ranges = list(c(0,16),c(0,30)),
isolines = list(list(x=c(0,16),y=c(0,0)),
list(x=c(0,2,10,16),y=c(5,5,15,15)),
list(x=c(0,16),y=c(15,15)),
list(x=c(0,16),y=c(30,30))),
u = c(0.0,0.6,1.0,1.0),
lead = 1,
utility = FALSE)
# define discrete end node for riparian zone vegetation
# (attriute "riparianzoneveg_class" with levels "natural",
# "seminatural" and "artificial")
riparzone_veg <-
utility.endnode.discrete.create(
name.node = "riparian zone veg.",
attrib.levels = data.frame(riparianzoneveg_class=
c("natural","seminatural","artificial")),
u = c(1,0.5625,0),
required = FALSE,
utility = FALSE)
# define aggregation node for riparian zone
riparzone <-
utility.aggregation.create(
name.node = "riparian zone",
nodes = list(riparzone_width,riparzone_veg),
name.fun = "utility.aggregate.cobbdouglas",
par = c(1,1),
required = FALSE)
# define aggregation node for ecomorphological state
morphol <-
utility.aggregation.create(
name.node = "ecomorphology",
nodes = list(widthvar,bedmod,bankmod,riparzone),
name.fun = "utility.aggregate.mix",
par = c(0.25,0.25,0.25,0.25,0,0,1),
names.par = c("w_widthvar","w_bedmod","w_bankmod","w_riparzone",
"w_add","w_min","w_cobbdouglas"),
required = TRUE)
# print individual definitions
print(widthvar)
print(bedmod)
# print all definitions
print(morphol)
# plot objectives hierarchy with attributes
plot(morphol)
# plot individual nodes:
plot(widthvar)
plot(widthvar,par=c(u_moderate=0.2))
plot(bedmod_other)
plot(bankmod)
#plot(riparzone_width) # too slow for package installation
# plot selected node definitions of a hierarchy
plot(morphol,type="nodes",nodes=c("width variability",
"bed modification other",
"bank modification"))
# evaluate value function for data sets and plot colored hierarchies
# and table
attrib_channelized <- data.frame(widthvariability_class = "none",
bedmodtype_class = "riprap",
bedmodfract_percent = 50,
bankmodtype_class = "imperm",
bankmodfract_percent = 70,
riverbedwidth_m = 10,
riparianzonewidth_m = 5,
riparianzoneveg_class = "seminatural")
attrib_rehab <- data.frame(widthvariability_class = "high",
bedmodtype_class = "riprap",
bedmodfract_percent = 50,
bankmodtype_class = "imperm",
bankmodfract_percent = 20,
riverbedwidth_m = 15,
riparianzonewidth_m = 15,
riparianzoneveg_class = "natural")
res_channelized <- evaluate(morphol,attrib=attrib_channelized)
res_channelized_add <- evaluate(morphol,attrib=attrib_channelized,
par=c(w_add=1,w_min=0,w_cobbdouglas=0))
res_rehab <- evaluate(morphol,attrib=attrib_rehab)
res_both <- rbind(res_channelized,res_rehab)
rownames(res_both) <- c("channelized","rehabilitated")
plot(morphol,u=res_channelized)
plot(morphol,u=res_channelized_add)
plot(morphol,u=res_rehab)
plot(morphol,u=res_rehab,uref=res_channelized)
plot(morphol,u=res_both,type="table",plot.val=FALSE)
plot(morphol,u=res_both,type="table",plot.val=TRUE,print.val=FALSE)
plot(morphol,u=res_both,uref=res_channelized,type="table",plot.val=FALSE)
# consideration of uncertain attribute levels
# (Higher uncertainty for predicted state after rehabilitation than for
# observed channelized state.
# Note that the normal distributions lead to a small probability of attribute
# levels beyond the range for which the value function is defined. This could
# be corrected for by truncating or choosing another distribution. We keep
# those values to demonstrate that this leads to warnings when evaluating the
# value function for these attribute levels,):
sampsize <- 1000
attrib_channelized_unc <- data.frame(
widthvariability_class = rep("high",sampsize),
bedmodtype_class = rep("riprap",sampsize),
bedmodfract_percent = rnorm(sampsize,mean=50,sd=5),
bankmodtype_class = rep("imperm",sampsize),
bankmodfract_percent = rnorm(sampsize,mean=70,sd=5),
riverbedwidth_m = rep(10,sampsize),
riparianzonewidth_m = rep(5,sampsize),
riparianzoneveg_class = c("seminatural","artificial")[rbinom(sampsize,1,0.5)+1])
attrib_rehab_unc <- data.frame(
widthvariability_class = c("moderate","high")[rbinom(sampsize,1,0.5)+1],
bedmodtype_class = rep("riprap",sampsize),
bedmodfract_percent = rnorm(sampsize,mean=50,sd=15),
bankmodtype_class = rep("imperm",sampsize),
bankmodfract_percent = rnorm(sampsize,mean=20,sd=5),
riverbedwidth_m = rnorm(sampsize,mean=10,sd=2),
riparianzonewidth_m = rnorm(sampsize,mean=10,sd=2),
riparianzoneveg_class = c("natural","seminatural")[rbinom(sampsize,1,0.5)+1])
res_channelized_unc <- evaluate(morphol,attrib=attrib_channelized_unc)
res_rehab_unc <- evaluate(morphol,attrib=attrib_rehab_unc)
plot(morphol,u=res_channelized_unc)
#plot(morphol,u=res_rehab_unc)
plot(morphol,u=res_rehab_unc,uref=res_channelized_unc)
plot(morphol,u=list(channelized=res_channelized_unc,rehabilitated=res_rehab_unc),
type="table")
plot(morphol,u=list(channelized=res_channelized_unc,rehabilitated=res_rehab_unc),
type="table",nodes=c("ecomorphology","riparian zone"))
plot(morphol,u=list(channelized=res_channelized_unc,rehabilitated=res_rehab_unc),
type="table",levels=2)
plot(morphol,u=list(channelized=res_channelized_unc,rehabilitated=res_rehab_unc),
uref=res_channelized_unc,
type="table")
utility documentation built on Aug. 28, 2023, 1:07 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
| utility-package | R Documentation |
Construct, Evaluate and Plot Value and Utility Functions
Description
Construct and plot objective hierarchies and associated value and utility functions. Evaluate the values and utilities and visualize the results as colored objective hierarchies or tables. Visualize uncertainty by plotting median and quantile intervals within the nodes of the objective hierarchy. Get numerical results of the evaluations in standard R data types for further processing.
Details
| Package: | utility |
| Type: | Package |
| Version: | 1.4.6 |
| Date: | 2023-08-27 |
| License: | GPL-3 |
An objective hierarchy and an associated value or utility function
is constructed by constructing the nodes of the hierarchy starting
from the end nodes and proceeding to the higher hierarchies.
Five types of end nodes are distinguished:
End nodes of the class utility.endnode.discrete define a value
or utility function for an attribute that has a finite number of
discrete numeric or non-numeric levels.
End nodes of the classes utility.endnode.intpol1d and
utility.endnode.parfun1d implement single-attribute
value or utility functions that accept a continuous argument.
The first of these functions allows the user to specify
attribute-value pairs and performs linear interpolation between these
points.
The second function allows the user to specify any parameteric
function that is implemented as a function in R.
End nodes of the class utility.endnode.intpol2d implement
interpolated value or utility functions that are based on two
attributes.
End nodes of the class utility.endnode.cond implement
value or utility functions that assign different value or utility
functions to a finite set of attribute combinations.
End nodes of the class utility.endnode.firstavail implement
value or utility functions that try to evaluate a list of nodes and
return the value of the first node that could be evaluated based
on the provided attribute data.
Finally, end nodes of the class utility.endnode.classcounts implement
value or utility functions that value counts e.g. of species of different classes
by assigning a basic value for the occurrence of at least one species of the best class
and incrementing this value by multiplicities of species of this class and of the next lower class.
These end nodes can be implemented by using the following constructors.
utility.endnode.discrete.create
utility.endnode.intpol1d.create
utility.endnode.parfun1d.create
utility.endnode.intpol2d.create
utility.endnode.cond.create
utility.endnode.firstavail.create
utility.endnode.classcounts.create
To advance to higher hierarchical levels, values or utilities at lower
levels must be aggregated to the next higher level.
This is done ab aggregation nodes of the class utility.aggregation.
Such nodes can be implemented by using the following constructor:
utility.aggregation.create
Finally, to provide decision support under uncertainty, values at
an adequate level of the objectives hierarchy must be converted to
utilities by accounting for the risk attitude of the decision maker.
Similar to the single-attribute value or utility functions,
this can either be done by linear interpolation with a node of the class
utility.conversion.intpol or by using a parametric funciton in
a node of the class utility.conversion.parfun.
These conversion nodes can be implemented by the constructors:
utility.conversion.intpol.create
utility.conversion.parfun.create
The definition of the objective hierarchy and the associated value
and utility function can then be listed or visualized by using the
generic functions
print
summary
plot
which automaticall call the implementation corresponding to the node specified
as the first argument:
print.utility.endnode.discrete
print.utility.endnode.intpol1d
print.utility.endnode.parfun1d
print.utility.endnode.intpol2d
print.utility.endnode.cond
print.utility.endnode.firstavail
print.utility.endnode.classcounts
print.utility.aggregation
print.utility.conversion.intpol
print.utility.conversion.parfun
summary.utility.endnode.discrete
summary.utility.endnode.intpol1d
summary.utility.endnode.parfun1d
summary.utility.endnode.intpol2d
summary.utility.endnode.cond
summary.utility.endnode.firstavail
summary.utility.endnode.classcounts
summary.utility.aggregation
summary.utility.conversion.intpol
summary.utility.conversion.parfun
plot.utility.endnode.discrete
plot.utility.endnode.intpol1d
plot.utility.endnode.parfun1d
plot.utility.endnode.intpol2d
plot.utility.endnode.cond
plot.utility.endnode.firstavail
plot.utility.endnode.classcounts
plot.utility.aggregation
plot.utility.conversion.intpol
plot.utility.conversion.parfun
The value or utility function can then be evaluated by applying the
generic function
evaluate
that again calls automatically the corresponding class-specific function
evaluate.utility.endnode.discrete
evaluate.utility.endnode.intpol1d
evaluate.utility.endnode.parfun1d
evaluate.utility.endnode.intpol2d
evaluate.utility.endnode.cond
evaluate.utility.endnode.firstavail
evaluate.utility.endnode.classcounts
evaluate.utility.aggregation
evaluate.utility.conversion.intpol
evaluate.utility.conversion.parfun
This function requires the provision of observed or predicted attributes
of the valued system and returns the corresponding values or utilities
of all nodes of the hierarchy.
These results can then be visualized by providing them to the generic
function
plot
in addition to the definition of the objective hierarchy stored in the
variable corresponding to the highest node of the hierarchy.
Again, this function automatically calls the correct class-specific
implementation (the root of the hierarchy will be an aggregation
or a conversion node, not an end node):
plot.utility.aggregation
plot.utility.conversion.intpol
plot.utility.conversion.parfun
This proceedure guarantees easy handling with the simple commands
print, summary, evaluate, and plot and
the specific function descriptions provided above are only required
to check advanced attributes.
Author(s)
Peter Reichert <peter.reichert@emeriti.eawag.ch> with contributions by Nele Schuwirth <nele.schuwirth@eawag.ch>
Maintainer: Peter Reichert <peter.reichert@emeriti.eawag.ch>
References
Short description of the package:
Reichert, P., Schuwirth, N. and Langhans, S.,
Constructing, evaluating and visualizing value and utility functions for decision support, Environmental Modelling & Software 46, 283-291, 2013.
Textbooks on the use of utility and value functions in decision analysis:
Keeney, R. L. and Raiffa, H. Decisions with Multiple Objectives - Preferences and Value Tradeoffs. John Wiley & Sons, 1976.
Eisenfuehr, F., Weber, M. and Langer, T., Rational Decision Making, Springer, Berlin, 2010.
Examples
# define discrete end node for width variability
# (attribute "widthvariability_class" with levels "high",
# "moderate" and "none")
widthvar <-
utility.endnode.discrete.create(
name.node = "width variability",
attrib.levels = data.frame(widthvariability_class=
c("high","moderate","none")),
u = c(1,0.4125,0),
names.u = c("u.high","u_moderate","u.none"),
required = FALSE,
utility = FALSE)
# define 1d interpolation end node for bed modification with
# riprap
# (attribute "bedmodfract_percent" with levels from 0 to 100)
bedmod_riprap <-
utility.endnode.intpol1d.create(
name.node = "bed modification riprap",
name.attrib = "bedmodfract_percent",
range = c(0,100),
x = c(0,10,30,100),
u = c(1,0.775,0.5625,0.24),
required = FALSE,
utility = FALSE)
# define 1d interpolation end node for bed modification with
# other material
# (attribute "bedmodfract_percent" with levels from 0 to 100)
bedmod_other <-
utility.endnode.intpol1d.create(
name.node = "bed modification other",
name.attrib = "bedmodfract_percent",
range = c(0,100),
x = c(0,10,30,100),
u = c(1,0.775,0.5625,0),
required = FALSE,
utility = FALSE)
# define combination end node for bed modification
# (attributes "bedmodtype_class" and "bedmodfract_percent")
bedmod <-
utility.endnode.cond.create(
name.node = "bed modification",
attrib.levels = data.frame(bedmodtype_class=
c("riprap","other")),
nodes = list(bedmod_riprap,bedmod_other),
required = FALSE,
utility = FALSE)
# define 1d interpolation end node for bank modification with
# permeable material
# (attribute "bankmodfract_percent" with levels from 0 to 100)
bankmod_perm <-
utility.endnode.intpol1d.create(
name.node = "bank modification perm",
name.attrib = "bankmodfract_percent",
range = c(0,100),
x = c(0,10,30,60,100),
u = c(1,0.8667,0.675,0.4125,0.24),
required = FALSE,
utility = FALSE)
# define 1d interpolation end node for bank modification with
# impermeable material
# (attribute "bankmodfract_percent" with levels from 0 to 100)
bankmod_imperm <-
utility.endnode.intpol1d.create(
name.node = "bank modification imperm",
name.attrib = "bankmodfract_percent",
range = c(0,100),
x = c(0,10,30,60,100),
u = c(1,0.775,0.5625,0.24,0),
required = FALSE,
utility = FALSE)
# define combination end node for bank modification
# (attributes "bankmodtype_class" and "bankmodfract_percent")
bankmod <-
utility.endnode.cond.create(
name.node = "bank modification",
attrib.levels = data.frame(bankmodtype_class=
c("perm","imperm")),
nodes = list(bankmod_perm,bankmod_imperm),
required = FALSE,
utility = FALSE)
# define 2d interpolation end node for riparian zone width
# (attributes "riparianzonewidth_m" and "riparianzonewidth_m")
riparzone_width <-
utility.endnode.intpol2d.create(
name.node = "riparian zone width",
name.attrib = c("riverbedwidth_m","riparianzonewidth_m"),
ranges = list(c(0,16),c(0,30)),
isolines = list(list(x=c(0,16),y=c(0,0)),
list(x=c(0,2,10,16),y=c(5,5,15,15)),
list(x=c(0,16),y=c(15,15)),
list(x=c(0,16),y=c(30,30))),
u = c(0.0,0.6,1.0,1.0),
lead = 1,
utility = FALSE)
# define discrete end node for riparian zone vegetation
# (attriute "riparianzoneveg_class" with levels "natural",
# "seminatural" and "artificial")
riparzone_veg <-
utility.endnode.discrete.create(
name.node = "riparian zone veg.",
attrib.levels = data.frame(riparianzoneveg_class=
c("natural","seminatural","artificial")),
u = c(1,0.5625,0),
required = FALSE,
utility = FALSE)
# define aggregation node for riparian zone
riparzone <-
utility.aggregation.create(
name.node = "riparian zone",
nodes = list(riparzone_width,riparzone_veg),
name.fun = "utility.aggregate.cobbdouglas",
par = c(1,1),
required = FALSE)
# define aggregation node for ecomorphological state
morphol <-
utility.aggregation.create(
name.node = "ecomorphology",
nodes = list(widthvar,bedmod,bankmod,riparzone),
name.fun = "utility.aggregate.mix",
par = c(0.25,0.25,0.25,0.25,0,0,1),
names.par = c("w_widthvar","w_bedmod","w_bankmod","w_riparzone",
"w_add","w_min","w_cobbdouglas"),
required = TRUE)
# print individual definitions
print(widthvar)
print(bedmod)
# print all definitions
print(morphol)
# plot objectives hierarchy with attributes
plot(morphol)
# plot individual nodes:
plot(widthvar)
plot(widthvar,par=c(u_moderate=0.2))
plot(bedmod_other)
plot(bankmod)
#plot(riparzone_width) # too slow for package installation
# plot selected node definitions of a hierarchy
plot(morphol,type="nodes",nodes=c("width variability",
"bed modification other",
"bank modification"))
# evaluate value function for data sets and plot colored hierarchies
# and table
attrib_channelized <- data.frame(widthvariability_class = "none",
bedmodtype_class = "riprap",
bedmodfract_percent = 50,
bankmodtype_class = "imperm",
bankmodfract_percent = 70,
riverbedwidth_m = 10,
riparianzonewidth_m = 5,
riparianzoneveg_class = "seminatural")
attrib_rehab <- data.frame(widthvariability_class = "high",
bedmodtype_class = "riprap",
bedmodfract_percent = 50,
bankmodtype_class = "imperm",
bankmodfract_percent = 20,
riverbedwidth_m = 15,
riparianzonewidth_m = 15,
riparianzoneveg_class = "natural")
res_channelized <- evaluate(morphol,attrib=attrib_channelized)
res_channelized_add <- evaluate(morphol,attrib=attrib_channelized,
par=c(w_add=1,w_min=0,w_cobbdouglas=0))
res_rehab <- evaluate(morphol,attrib=attrib_rehab)
res_both <- rbind(res_channelized,res_rehab)
rownames(res_both) <- c("channelized","rehabilitated")
plot(morphol,u=res_channelized)
plot(morphol,u=res_channelized_add)
plot(morphol,u=res_rehab)
plot(morphol,u=res_rehab,uref=res_channelized)
plot(morphol,u=res_both,type="table",plot.val=FALSE)
plot(morphol,u=res_both,type="table",plot.val=TRUE,print.val=FALSE)
plot(morphol,u=res_both,uref=res_channelized,type="table",plot.val=FALSE)
# consideration of uncertain attribute levels
# (Higher uncertainty for predicted state after rehabilitation than for
# observed channelized state.
# Note that the normal distributions lead to a small probability of attribute
# levels beyond the range for which the value function is defined. This could
# be corrected for by truncating or choosing another distribution. We keep
# those values to demonstrate that this leads to warnings when evaluating the
# value function for these attribute levels,):
sampsize <- 1000
attrib_channelized_unc <- data.frame(
widthvariability_class = rep("high",sampsize),
bedmodtype_class = rep("riprap",sampsize),
bedmodfract_percent = rnorm(sampsize,mean=50,sd=5),
bankmodtype_class = rep("imperm",sampsize),
bankmodfract_percent = rnorm(sampsize,mean=70,sd=5),
riverbedwidth_m = rep(10,sampsize),
riparianzonewidth_m = rep(5,sampsize),
riparianzoneveg_class = c("seminatural","artificial")[rbinom(sampsize,1,0.5)+1])
attrib_rehab_unc <- data.frame(
widthvariability_class = c("moderate","high")[rbinom(sampsize,1,0.5)+1],
bedmodtype_class = rep("riprap",sampsize),
bedmodfract_percent = rnorm(sampsize,mean=50,sd=15),
bankmodtype_class = rep("imperm",sampsize),
bankmodfract_percent = rnorm(sampsize,mean=20,sd=5),
riverbedwidth_m = rnorm(sampsize,mean=10,sd=2),
riparianzonewidth_m = rnorm(sampsize,mean=10,sd=2),
riparianzoneveg_class = c("natural","seminatural")[rbinom(sampsize,1,0.5)+1])
res_channelized_unc <- evaluate(morphol,attrib=attrib_channelized_unc)
res_rehab_unc <- evaluate(morphol,attrib=attrib_rehab_unc)
plot(morphol,u=res_channelized_unc)
#plot(morphol,u=res_rehab_unc)
plot(morphol,u=res_rehab_unc,uref=res_channelized_unc)
plot(morphol,u=list(channelized=res_channelized_unc,rehabilitated=res_rehab_unc),
type="table")
plot(morphol,u=list(channelized=res_channelized_unc,rehabilitated=res_rehab_unc),
type="table",nodes=c("ecomorphology","riparian zone"))
plot(morphol,u=list(channelized=res_channelized_unc,rehabilitated=res_rehab_unc),
type="table",levels=2)
plot(morphol,u=list(channelized=res_channelized_unc,rehabilitated=res_rehab_unc),
uref=res_channelized_unc,
type="table")
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.