## ExperienceDecay
## 1 TRUE
## 2 FALSE
Cover of subscenarios and sd
## [1] 200
## [1] 0
Number of results per subscenario
## [1] 400
Box plots of time execution
library(ggplot2)
# Boxplots of task count by scenario number
# observations (points) are overlayed and jittered
qplot(OptionGroup, Tick_Count, data=batch_merged, geom=c("boxplot", "jitter"),
fill=OptionGroup, main="Tick count per different scenario",
xlab="", ylab="Number of ticks") + stat_boxplot(geom ='errorbar')
Table with winning strategies is produced here:
library(xtable)
print(xtable(winning_p7), type = "html", include.rownames = T)
OptionGroup | homophily | heterophily | preferential | hom_het | hom_pre | het_pre | hom_het_pre | |
---|---|---|---|---|---|---|---|---|
1 | ExpDec-0 | 200 | 0 | 0 | 0 | 0 | 0 | 0 |
2 | ExpDec-1 | 200 | 0 | 0 | 0 | 0 | 0 | 0 |
Winning strategy is a strategy which all agents took consequently during 10 generations. In case of more than one strategy, we have a “mix” of winning strategies which stays constant during 10 generations.
Each row represent a single subscenario, with a count of winning strategies per strategy type (hence this value means “how many times in different runs this scenario dominated”, it's not agent's count or any other value).
## ExperienceDecay
## 1 TRUE
## 2 FALSE
Cover of subscenarios and sd
## [1] 200
## [1] 0
Number of results per subscenario
## [1] 400
Box plots of time execution
library(ggplot2)
# Boxplots of task count by scenario number
# observations (points) are overlayed and jittered
qplot(OptionGroup, Tick_Count, data=batch_merged, geom=c("boxplot", "jitter"),
fill=OptionGroup, main="Tick count per different scenario",
xlab="", ylab="Number of ticks") + stat_boxplot(geom ='errorbar')
Table with winning strategies is produced here:
library(xtable)
print(xtable(winning_p6), type = "html", include.rownames = T)
OptionGroup | homophily | heterophily | preferential | hom_het | hom_pre | het_pre | hom_het_pre | |
---|---|---|---|---|---|---|---|---|
1 | ExpDec-0 | 0 | 182 | 17 | 0 | 0 | 1 | 0 |
2 | ExpDec-1 | 0 | 41 | 156 | 0 | 0 | 3 | 0 |
Winning strategy is a strategy which all agents took consequently during 10 generations. In case of more than one strategy, we have a “mix” of winning strategies which stays constant during 10 generations.
Each row represent a single subscenario, with a count of winning strategies per strategy type (hence this value means “how many times in different runs this scenario dominated”, it's not agent's count or any other value).
## ExperienceDecay
## 1 TRUE
## 2 FALSE
Cover of subscenarios and sd
## [1] 200
## [1] 0
Number of results per subscenario
## [1] 400
Box plots of time execution
library(ggplot2)
# Boxplots of task count by scenario number
# observations (points) are overlayed and jittered
qplot(OptionGroup, Tick_Count, data=batch_merged, geom=c("boxplot", "jitter"),
fill=OptionGroup, main="Tick count per different scenario",
xlab="", ylab="Number of ticks") + stat_boxplot(geom ='errorbar')
Table with winning strategies is produced here:
library(xtable)
print(xtable(winning_p5), type = "html", include.rownames = T)
OptionGroup | homophily | heterophily | preferential | hom_het | hom_pre | het_pre | hom_het_pre | |
---|---|---|---|---|---|---|---|---|
1 | ExpDec-0 | 200 | 0 | 0 | 0 | 0 | 0 | 0 |
2 | ExpDec-1 | 200 | 0 | 0 | 0 | 0 | 0 | 0 |
Winning strategy is a strategy which all agents took consequently during 10 generations. In case of more than one strategy, we have a “mix” of winning strategies which stays constant during 10 generations.
Each row represent a single subscenario, with a count of winning strategies per strategy type (hence this value means “how many times in different runs this scenario dominated”, it's not agent's count or any other value).
## ExperienceDecay
## 1 TRUE
## 2 FALSE
Cover of subscenarios and sd
## [1] 200
## [1] 0
Number of results per subscenario
## [1] 400
Box plots of time execution
library(ggplot2)
# Boxplots of task count by scenario number
# observations (points) are overlayed and jittered
qplot(OptionGroup, Tick_Count, data=batch_merged, geom=c("boxplot", "jitter"),
fill=OptionGroup, main="Tick count per different scenario",
xlab="", ylab="Number of ticks") + stat_boxplot(geom ='errorbar')
Table with winning strategies is produced here:
library(xtable)
print(xtable(winning_p4), type = "html", include.rownames = T)
OptionGroup | homophily | heterophily | preferential | hom_het | hom_pre | het_pre | hom_het_pre | |
---|---|---|---|---|---|---|---|---|
1 | ExpDec-0 | 0 | 173 | 27 | 0 | 0 | 0 | 0 |
2 | ExpDec-1 | 0 | 96 | 104 | 0 | 0 | 0 | 0 |
Winning strategy is a strategy which all agents took consequently during 10 generations. In case of more than one strategy, we have a “mix” of winning strategies which stays constant during 10 generations.
Each row represent a single subscenario, with a count of winning strategies per strategy type (hence this value means “how many times in different runs this scenario dominated”, it's not agent's count or any other value).
## ExperienceDecay
## 1 TRUE
## 2 FALSE
Cover of subscenarios and sd
## [1] 200
## [1] 0
Number of results per subscenario
## [1] 400
Box plots of time execution
library(ggplot2)
# Boxplots of task count by scenario number
# observations (points) are overlayed and jittered
qplot(OptionGroup, Tick_Count, data=batch_merged, geom=c("boxplot", "jitter"),
fill=OptionGroup, main="Tick count per different scenario",
xlab="", ylab="Number of ticks") + stat_boxplot(geom ='errorbar')
Table with winning strategies is produced here:
library(xtable)
print(xtable(winning_p3), type = "html", include.rownames = T)
OptionGroup | homophily | heterophily | preferential | hom_het | hom_pre | het_pre | hom_het_pre | |
---|---|---|---|---|---|---|---|---|
1 | ExpDec-0 | 200 | 0 | 0 | 0 | 0 | 0 | 0 |
2 | ExpDec-1 | 200 | 0 | 0 | 0 | 0 | 0 | 0 |
Winning strategy is a strategy which all agents took consequently during 10 generations. In case of more than one strategy, we have a “mix” of winning strategies which stays constant during 10 generations.
Each row represent a single subscenario, with a count of winning strategies per strategy type (hence this value means “how many times in different runs this scenario dominated”, it's not agent's count or any other value).
## ExperienceDecay
## 1 TRUE
## 2 FALSE
Cover of subscenarios and sd
## [1] 200
## [1] 31.81981
Number of results per subscenario
## [1] 355
Box plots of time execution
library(ggplot2)
# Boxplots of task count by scenario number
# observations (points) are overlayed and jittered
qplot(OptionGroup, Tick_Count, data=batch_merged, geom=c("boxplot", "jitter"),
fill=OptionGroup, main="Tick count per different scenario",
xlab="", ylab="Number of ticks") + stat_boxplot(geom ='errorbar')
Table with winning strategies is produced here:
library(xtable)
print(xtable(winning_p2), type = "html", include.rownames = T)
OptionGroup | homophily | heterophily | preferential | hom_het | hom_pre | het_pre | hom_het_pre | |
---|---|---|---|---|---|---|---|---|
1 | ExpDec-0 | 200 | 0 | 0 | 0 | 0 | 0 | 0 |
2 | ExpDec-1 | 155 | 0 | 0 | 0 | 0 | 0 | 0 |
Winning strategy is a strategy which all agents took consequently during 10 generations. In case of more than one strategy, we have a “mix” of winning strategies which stays constant during 10 generations.
Each row represent a single subscenario, with a count of winning strategies per strategy type (hence this value means “how many times in different runs this scenario dominated”, it's not agent's count or any other value).
## ExperienceDecay
## 1 TRUE
## 2 FALSE
Cover of subscenarios and sd
## [1] 143
## [1] 16.26346
Number of results per subscenario
## [1] 263
Box plots of time execution
library(ggplot2)
# Boxplots of task count by scenario number
# observations (points) are overlayed and jittered
qplot(OptionGroup, Tick_Count, data=batch_merged, geom=c("boxplot", "jitter"),
fill=OptionGroup, main="Tick count per different scenario",
xlab="", ylab="Number of ticks") + stat_boxplot(geom ='errorbar')
Table with winning strategies is produced here:
library(xtable)
print(xtable(winning_p1), type = "html", include.rownames = T)
OptionGroup | homophily | heterophily | preferential | hom_het | hom_pre | het_pre | hom_het_pre | |
---|---|---|---|---|---|---|---|---|
1 | ExpDec-0 | 142 | 0 | 0 | 1 | 0 | 0 | 0 |
2 | ExpDec-1 | 120 | 0 | 0 | 0 | 0 | 0 | 0 |
Winning strategy is a strategy which all agents took consequently during 10 generations. In case of more than one strategy, we have a “mix” of winning strategies which stays constant during 10 generations.
Each row represent a single subscenario, with a count of winning strategies per strategy type (hence this value means “how many times in different runs this scenario dominated”, it's not agent's count or any other value).
Experience decay TRUE
matrix_expd_TRUE
## homophily heterophily preferential
## homophily -1 5 5
## heterophily 0 -1 0
## preferential 2 2 -1
Experience decay FALSE
matrix_expd_FALSE
## homophily heterophily preferential
## homophily -1 5 5
## heterophily 2 -1 2
## preferential 0 0 -1
Game winners / loosers on heatmaps
matrix_expd_TRUE[matrix_expd_TRUE == -1] <- NA
matrix_expd_FALSE[matrix_expd_FALSE == -1] <- NA
melted_avg_expd1 <- melt(matrix_expd_TRUE)
melted_avg_expd0 <- melt(matrix_expd_FALSE)
colnames(melted_avg_expd1) <- c("winning", "looses", "times")
colnames(melted_avg_expd0) <- c("winning", "looses", "times")
p_avg_expd1 <- ggplot(melted_avg_expd1, aes(looses, winning)) +
geom_tile(aes(fill = cut(times, breaks=-1:5, labels=0:5)), colour = "grey") +
scale_fill_manual(drop=FALSE, values=colorRampPalette(c("white","red"))(6), na.value="#EAEAEA", name="Times") +
xlab("looses against") + ylab("winning over") +
ggtitle("Avg() function, experience decay ON") +
scale_y_discrete(limits=c("preferential", "heterophily", "homophily"))
p_avg_expd0 <- ggplot(melted_avg_expd0, aes(looses, winning)) +
geom_tile(aes(fill = cut(times, breaks=-1:5, labels=0:5)), colour = "grey") +
scale_fill_manual(drop=FALSE, values=colorRampPalette(c("white","red"))(6), na.value="#EAEAEA", name="Times") +
xlab("looses against") + ylab("winning over") +
ggtitle("Avg() function, experience decay OFF") +
scale_y_discrete(limits=c("preferential", "heterophily", "homophily"))
multiplot(p_avg_expd1, p_avg_expd0, cols=1)