| Title: | Using Percolation and Conductance to Find Information Flow Certainty in a Direct Network |
|---|---|
| Description: | To find the certainty of dominance interactions with indirect interactions being considered. |
| Authors: | Kevin Fujii [aut], Jian Jin [aut], Jessica Vandeleest [aut, cre], Aaron Shev [aut], Brianne Beisner [aut], Brenda McCowan [aut, cph], Hsieh Fushing [aut, cph] |
| Maintainer: | Jessica Vandeleest <[email protected]> |
| License: | GPL (>= 2) |
| Version: | 0.1.6 |
| Built: | 2025-03-04 05:18:46 UTC |
| Source: | https://github.com/hanettools/perc |
conf.mat classas.conflictmat convert an edgelist or a win-loss raw matrix to a matrix of conf.mat class
as.conflictmat(Data, weighted = FALSE, swap.order = FALSE)as.conflictmat(Data, weighted = FALSE, swap.order = FALSE)
Data |
either a dataframe or a matrix, representing raw win-loss interactions using either an edgelist or a matrix. By default, winners are represented by IDs in the 1st column for an edgelist, and by row IDs for a matrix. Frequency of interactions for each dyad can be represented either by multiple occurrences of the dyad for a 2-column edgelist, or by a third column specifying the frequency of the interaction for a 3-column edgelist. |
weighted |
If the edgelist is a 3-column edgelist in which weight was specified by frequency, use |
swap.order |
If the winner is placed in the 2nd column for an edgelist or as the column name for a matrix, specify as |
conf.mat is short for "Conflict Matrix". conf.mat is
a class of R objects. It is required to use as.conflictmat to convert your
raw edgelist or raw win-loss matrix into a matrix of conf.mat object before
using other functions to find (in)direct pathways and computing dominance probabilities.
Note, when using a 3-column edgelist (e.g. a weighted edgelist) to represent raw win-loss interactions, each dyad must be unique. If more than one rows are found with the same initiator and recipient, sum of the frequencies will be taken to represent the freqency of interactions between this unique dyad. A warning message will prompt your attention to the accuracy of your raw data when duplicate dyads were found in a three-column edgelist.
a named matrix with the [i,j]th entry equal to the number of times i wins over j.
findIDpaths, countPaths, transitivity, conductance
confmatrix <- as.conflictmat(sampleEdgelist, swap.order = FALSE) confmatrix2 <- as.conflictmat(sampleRawMatrix, swap.order = FALSE) confmatrix3 <- as.conflictmat(sampleWeightedEdgelist, weighted = TRUE, swap.order = FALSE)confmatrix <- as.conflictmat(sampleEdgelist, swap.order = FALSE) confmatrix2 <- as.conflictmat(sampleRawMatrix, swap.order = FALSE) confmatrix3 <- as.conflictmat(sampleWeightedEdgelist, weighted = TRUE, swap.order = FALSE)
bradleyTerry Computes the MLE for the BT model using an MM algorithm
bradleyTerry(conf.mat, initial = NA, baseline = NA, stop.dif = 0.001)bradleyTerry(conf.mat, initial = NA, baseline = NA, stop.dif = 0.001)
conf.mat |
a matrix of conf.mat class. An N-by-N conflict matrix whose |
initial |
initial values of dominance indices for the MM algorithm, if not supplied, the 0 vector will be the inital value. |
baseline |
index for agent to represent baseline dominance index set to 0. If NA, the "sum-to-one" parameterization will be used. |
stop.dif |
numeric value for difference in log likelihood value between iterations. Used as the convergence criterion for the algorithm. |
In order to meet Bradley-Terry assumption, each ID in conf.mat should have at least one win AND one loss.
bradleyTerry will return an error if no more than one win or loss was found.
@references Shev, A., Hsieh, F., Beisner, B., & McCowan, B. (2012). Using Markov chain Monte Carlo (MCMC) to visualize and test the linearity assumption of the Bradley-Terry class of models. Animal behaviour, 84(6), 1523-1531.
Shev, A., Fujii, K., Hsieh, F., & McCowan, B. (2014). Systemic Testing on Bradley-Terry Model against Nonlinear Ranking Hierarchy. PloS one, 9(12), e115367.
A list of length 3.
domInds |
a vector of length N consiting of the MLE values of the dominance indices. Lower values represent lower ranks. |
probMat |
an N-by-N numeric matrix of win-loss probabilities estimated by the BT model. |
logLik |
the model fit. |
# create an edgelist edgelist1 <- data.frame(col1 = sample(letters[1:15], 200, replace = TRUE), col2 = sample(letters[1:15], 200, replace = TRUE), stringsAsFactors = FALSE) edgelist1 <- edgelist1[-which(edgelist1$col1 == edgelist1$col2), ] # convert an edgelist to conflict matrix confmatrix_bt <- as.conflictmat(edgelist1) # Computes the MLE for the BT model bt <- bradleyTerry(confmatrix_bt)# create an edgelist edgelist1 <- data.frame(col1 = sample(letters[1:15], 200, replace = TRUE), col2 = sample(letters[1:15], 200, replace = TRUE), stringsAsFactors = FALSE) edgelist1 <- edgelist1[-which(edgelist1$col1 == edgelist1$col2), ] # convert an edgelist to conflict matrix confmatrix_bt <- as.conflictmat(edgelist1) # Computes the MLE for the BT model bt <- bradleyTerry(confmatrix_bt)
bt.test Systemic test for the assumptions of the Bradley-Terry model,
transitivity and monotonic win-loss relationship.
That is, if and then and
> .
bt.test(conf.mat, baseline = 1, maxLength = 2, reps = 1000)bt.test(conf.mat, baseline = 1, maxLength = 2, reps = 1000)
conf.mat |
an N-by-N matrix. The matrix should be a conflict matrix with element i,j representing the number of times i has beaten j. |
baseline |
an integer between 1 and N inclusive identifying the agent with dominance index equal to zero. |
maxLength |
an integer indicating maximum path length used
in |
reps |
an integer indicating number of conflict matrices simulated to estimate the sampling distribution under the BT model. |
The value of the test statistic should be within the estimated sampling distribution of the test statistics under the BT model. The p-value of the test indicates the probability of statistics in the estimated sampling distribution is larger than the test statistic. It is not appropriate to use Bradley-Terry model if value of the test statistic is higher than the estimated sampling distribution of the test statistics.
A list of 3 elements.
stat |
value of the test statistic |
dist |
estimated sampling distribution of the test statistics under the BT model. |
p.val |
p-value of the test |
Shev, A., Fujii, K., Hsieh, F., & McCowan, B. (2014). Systemic Testing on Bradley-Terry Model against Nonlinear Ranking Hierarchy. PloS one, 9(12), e115367.
# create an edgelist edgelist1 <- data.frame(col1 = sample(letters[1:15], 200, replace = TRUE), col2 = sample(letters[1:15], 200, replace = TRUE), stringsAsFactors = FALSE) edgelist1 <- edgelist1[-which(edgelist1$col1 == edgelist1$col2), ] # convert an edgelist to conflict matrix confmatrix_bt <- as.conflictmat(edgelist1) # test the assumptions of the Bradley-Terry model # not run: # condTestoutput <- bt.test(confmatrix_bt)# create an edgelist edgelist1 <- data.frame(col1 = sample(letters[1:15], 200, replace = TRUE), col2 = sample(letters[1:15], 200, replace = TRUE), stringsAsFactors = FALSE) edgelist1 <- edgelist1[-which(edgelist1$col1 == edgelist1$col2), ] # convert an edgelist to conflict matrix confmatrix_bt <- as.conflictmat(edgelist1) # test the assumptions of the Bradley-Terry model # not run: # condTestoutput <- bt.test(confmatrix_bt)
conductance compute win-loss probabilities for all possible pairs
based upon the combined information from directed wins/losses and
indirect win/loss pathways from the network.
conductance(conf, maxLength, alpha = NULL, beta = 1, strict = FALSE)conductance(conf, maxLength, alpha = NULL, beta = 1, strict = FALSE)
conf |
a matrix of conf.mat class. An N-by-N conflict matrix whose |
maxLength |
an integer greater than 1 and less than 7, indicating the maximum length of paths to identify. |
alpha |
a positive integer that
reflects the influence of an observed win/loss interaction
on an underlying win-loss probability.
It is used in the calculation of the posterior distribution
for the win-loss probability of |
beta |
a positive numeric value that, like alpha,
reflects the influence of an observed win/loss interaction
on an underlying win-loss probability.
Both |
strict |
a logical vector of length 1. It is used in transitivity definition for alpha estimation. It should be set to TRUE when a transitive triangle is defined as all pathways in the triangle go to the same direction; it should be set to FALSE when a transitive triangle is defined as PRIMARY pathways in the triangle go to the same direction. Strict = FALSE by default. |
This function performs two major steps.
First, repeated random walks through the empirical network
identify all possible directed win-loss pathways
between each pair of nodes in the network.
Second, the information from both direct wins/losses and
pathways of win/loss interactions are combined into an estimate of
the underlying probability of i over j, for all ij pairs.
a list of two elements.
imputed.conf |
An N-by-N conflict matrix whose |
p.hat |
An N-by-N numeric matrix whose |
Fushing H, McAssey M, Beisner BA, McCowan B. 2011. Ranking network of a captive rhesus macaque society: a sophisticated corporative kingdom. PLoS ONE 6(3):e17817.
as.conflictmat, findIDpaths, transitivity, simRankOrder
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2, strict = FALSE) perm2$imputed.conf perm2$p.hat# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2, strict = FALSE) perm2$imputed.conf perm2$p.hat
countPaths Identifies the number of paths of length
less than or equal to maxLength between all pairs
countPaths(conf, maxLength = 2)countPaths(conf, maxLength = 2)
conf |
a matrix of conf.mat class.
An N-by-N conflict matrix whose |
maxLength |
a positive numeric integer indicating the maximum length of paths to identify |
A list in which elements are number of paths between all pairs of a given length.
as.conflictmat, findIDpaths, transitivity, conductance
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find number of paths of length 3 or less npaths <- countPaths(confmatrix, 3)# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find number of paths of length 3 or less npaths <- countPaths(confmatrix, 3)
dyadicLongConverter convert win-loss probability matrix into long format for each dyad
dyadicLongConverter(matrix)dyadicLongConverter(matrix)
matrix |
the win-loss matrix which is the second output from |
values on the diagonal of the matrix are not included in the converted long-format data.
a dataframe of dyadic level win-loss probability and ranking certainty.
conductance, valueConverter, individualDomProb
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2) perm2$imputed.conf perm2$p.hat dl <- dyadicLongConverter(perm2$p.hat)# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2) perm2$imputed.conf perm2$p.hat dl <- dyadicLongConverter(perm2$p.hat)
findAllPaths Identifies all paths length less than or equal
to maxLength between all pairs of competitors
findAllPaths(conf, maxLength = 2)findAllPaths(conf, maxLength = 2)
conf |
a matrix of conf.mat class. An N-by-N conflict matrix whose |
maxLength |
a positive numeric integer indicating the maximum length of paths to identify |
A list of two elements.
direct pathways |
direct pathways found in original matrix |
indirect pathways |
a list of all paths from length 2 to the given length |
countPaths findIDpaths transitivity
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find all paths of legnth 3 allp.3 <- findAllPaths(confmatrix, 3)# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find all paths of legnth 3 allp.3 <- findAllPaths(confmatrix, 3)
findIDpaths identifies all unique win-loss pathways of order beginning at selected ID
findIDpaths(conf, ID, len = 2)findIDpaths(conf, ID, len = 2)
conf |
a matrix of conf.mat class. An N-by-N conflict matrix whose |
ID |
a numeric or character vector of length 1. It specifys the subject at the beginning of each pathway. |
len |
a positive integer of length 1 greater than 2. the length of the win-loss paths to be identified ( |
return all win-loss paths of length(len) beginning at ID
as.conflictmat, findAllPaths, countPaths
confmatrix <- as.conflictmat(sampleEdgelist) path38891 <- findIDpaths(confmatrix, ID = "Kuai", len = 2)confmatrix <- as.conflictmat(sampleEdgelist) path38891 <- findIDpaths(confmatrix, ID = "Kuai", len = 2)
Associate each costs with its corresponding simulated annealing runs
getAllCosts(costs_all, num)getAllCosts(costs_all, num)
costs_all |
costs of all simAnnealing runs. It is the first element of the output from |
num |
number of simulated annealing runs |
a data.frame of all costs.
assign IDs to all best rank orders
getAllRankOrder(ID_index, allRankOrder)getAllRankOrder(ID_index, allRankOrder)
ID_index |
it depends on the inputed data from simRankOrder. It takes the colnames of data as ID, and index this ID by its position in the colname. |
allRankOrder |
all rank orders found in all simulated annealing runs. It is the third output from |
a data.frame of all costs.
assign IDs to the best rank order
getBestRankOrder(ID_index, bestRankOrder)getBestRankOrder(ID_index, bestRankOrder)
ID_index |
it depends on the inputed data from simRankOrder. It takes the colnames of data as ID, and index this ID by its position in the colname. |
bestRankOrder |
the best rank order found in all simulated annealing runs. It is the second output from |
a data.frame of all costs.
get useful outputs from simulated annealing processes
getSimOutput(simAnnealList, num)getSimOutput(simAnnealList, num)
simAnnealList |
the output from simAnnealing process |
num |
number of simulated annealing runs |
a list of three elements
costs_all |
costs of all simulated annealing runs. |
bestRankOrder |
best rank order found in all simulated annealing processes |
allRankOrder |
a dataframe, all best rank orders found in each simulated annealing processes |
individualDomProb convert win-loss probability matrix into long format for each dyad
individualDomProb(matrix)individualDomProb(matrix)
matrix |
the win-loss matrix which is the second output from |
a dataframe. Averaging probability of win-loss relationship with all other individuals.
conductance, valueConverter, dyadicLongConverter
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2) perm2$imputed.conf perm2$p.hat individualLevelOutput <- individualDomProb(perm2$p.hat)# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2) perm2$imputed.conf perm2$p.hat individualLevelOutput <- individualDomProb(perm2$p.hat)
plotConfmat generate heat map for a matrix or a win-loss probability matrix
plotConfmat(conf.mat, ordering = NA, labels = FALSE, ...)plotConfmat(conf.mat, ordering = NA, labels = FALSE, ...)
conf.mat |
an N-by-N matrix. Either a conflict matrix or
a win-loss probability matrix (the second element from |
ordering |
a reordering of the rows/columns, specified by a permutation of 1:N |
labels |
if TRUE, displaying the agent names as
specified in the |
... |
Further argument may be supplied and processed by |
A heatmap
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2) # plotting plotConfmat(perm2$p.hat)# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2) # plotting plotConfmat(perm2$p.hat)
plotProbDiagnosis generate heat map for dominance probability matrixDiagnosis Plot
plotProbDiagnosis generate heat map for dominance probability matrix
plotProbDiagnosis(prob.mat, cutoff = 0.75, ...)plotProbDiagnosis(prob.mat, cutoff = 0.75, ...)
prob.mat |
dominance probability matrix |
cutoff |
a numeric value between 0.5 to 1. A value that is equal or greater than the cutoff is considered of high certainty. |
... |
Further argument may be supplied and processed by |
sampleEdgelist. social interactions among 11 monkeys
sampleEdgelistsampleEdgelist
A data frame of edgelist with 174 rows and 2 variables: Iname, Rname
winner, animal ID
loser, animal ID
... McCowan Lab sample data.
sampleRawMatrix. dominance interactions between 39 monkeys
sampleRawMatrixsampleRawMatrix
A 39 x 39 matrix representing number of times that a row wins over a column
McCowan Lab sample data.
sampleWeightedEdgelist. dominance interactions among 29 monkeys
sampleWeightedEdgelistsampleWeightedEdgelist
A data frame of edgelist with 181 rows and 3 variables: Initiator1, Recipient1, Freq
winner, monkey name
loser, monkey name
Frequency, count of interaction
... McCowan Lab sample data.
simRankOrder find the rank order for the win-loss relationship
simRankOrder(data, num = 10, alpha = NULL, kmax = 1000)simRankOrder(data, num = 10, alpha = NULL, kmax = 1000)
data |
a matrix. the win-loss probability matrix
which is the second element of the output from |
num |
number of SimAnnealing (default is set at 10) |
alpha |
a positive integer that
reflects the influence of an observed win/loss interaction
on an underlying win-loss probability.
It is used in the calculation of the posterior distribution
for the win-loss probability of |
kmax |
an integer between 2 to 1000, indicating the number of simulations in each SimAnnealing. |
a list of two dataframes.
BestSimulatedRankOrder |
a dataframe representing the best simulated rank order. |
Costs |
the cost of each simulated annealing run |
AllSimulatedRankOrder |
a dataframe representing all simulated rank orders. |
Fushing, H., McAssey, M. P., Beisner, B., & McCowan, B. (2011). Ranking network of a captive rhesus macaque society: a sophisticated corporative kingdom. PLoS One, 6(3), e17817-e17817.
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find dominance probability matrix perm2 <- conductance(confmatrix, maxLength = 2) ## Not run: # Note: It takes a while to run the simRankOrder example. s.rank <- simRankOrder(perm2$p.hat, num = 10, kmax = 1000) s.rank$BestSimulatedRankOrder s.rank$Costs s.rank$AllSimulatedRankOrder ## End(Not run)# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find dominance probability matrix perm2 <- conductance(confmatrix, maxLength = 2) ## Not run: # Note: It takes a while to run the simRankOrder example. s.rank <- simRankOrder(perm2$p.hat, num = 10, kmax = 1000) s.rank$BestSimulatedRankOrder s.rank$Costs s.rank$AllSimulatedRankOrder ## End(Not run)
transitivity calculate transitivity measurements for a matrix
transitivity(conf, strict = FALSE)transitivity(conf, strict = FALSE)
conf |
an N-by-N conflict matrix whose |
strict |
a logical vector of length 1 (TRUE or FALSE). It is used in transitivity definition for alpha estimation. It should be set to TRUE when a transitive triangle is defined as all pathways in the triangle go to the same direction; it should be set to FALSE when a transitive triangle is defined as PRIMARY pathways in the triangle go to the same direction. Strict = FALSE by default. |
transitivity is calculated as the proportion transitive triangles in the total of transitive and intransitive triangles.
transitivity is used to estimate alpha, which is used in turn in imputing information from indirect pathways as to what degree we can trust information from indirect pathways.
Greater transitivity is associated with assigning higher weight to information from indirect pathways.
A list of four elements.
transitive |
The number of transitive triangles. |
intransitive |
The number of intransitive triangles. |
transitivity |
transitivity, the proportion of transitive triangles. |
alpha |
The value of alpha corresponding to this value of transitivity. |
countPaths, findIDpaths, conductance
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # transitivity calculation conftrans <- transitivity(confmatrix, strict = FALSE) conftrans$transitive conftrans$intransitive conftrans$transitivity conftrans$alpha# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # transitivity calculation conftrans <- transitivity(confmatrix, strict = FALSE) conftrans$transitive conftrans$intransitive conftrans$transitivity conftrans$alpha
valueConverter converts or transforms all values (which range from 0.0 to 1.0)
in the win-loss probability matrix into 0.5 - 1.0
valueConverter(matrix)valueConverter(matrix)
matrix |
the win-loss matrix which is the second output from |
a matrix of win-loss probability ranging from 0.5 - 1.0.
conductance, individualDomProb, dyadicLongConverter
# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2) perm2$imputed.conf perm2$p.hat convertedValue <- valueConverter(perm2$p.hat)# convert an edgelist to conflict matrix confmatrix <- as.conflictmat(sampleEdgelist) # find win-loss probability matrix perm2 <- conductance(confmatrix, 2) perm2$imputed.conf perm2$p.hat convertedValue <- valueConverter(perm2$p.hat)