Feature selection comparison using simulated data from an erdos-renyi graph structure
This commit is contained in:
parent
c8cac4a638
commit
548b426e84
@ -14,156 +14,94 @@ source("Schrick-Noah_Ridge-LASSO-Regression.R")
|
|||||||
source("Schrick-Noah_Simulated-Data.R")
|
source("Schrick-Noah_Simulated-Data.R")
|
||||||
bundled_data <- create_data()
|
bundled_data <- create_data()
|
||||||
|
|
||||||
### LASSO
|
run_comparison <- function(bundled_data){
|
||||||
unpen_beta <- unpen_coeff(bundled_data$train.X, bundled_data$train.y)
|
### LASSO
|
||||||
lasso.df <- data.frame(att=c("intercept", colnames(bundled_data$train.X)),
|
unpen_beta <- unpen_coeff(bundled_data$train.X, bundled_data$train.y)
|
||||||
|
lasso.df <- data.frame(att=c("intercept", colnames(bundled_data$train.X)),
|
||||||
scores=unpen_beta$betas,
|
scores=unpen_beta$betas,
|
||||||
abs_scores=abs(unpen_beta$betas))
|
abs_scores=abs(unpen_beta$betas))
|
||||||
lasso.res <- dplyr::slice_max(lasso.df,order_by=abs_scores,n=20)
|
lasso.res <- dplyr::slice_max(lasso.df,order_by=abs_scores,n=20)
|
||||||
lasso.table <- as.data.table(lasso.res)
|
lasso.table <- as.data.table(lasso.res)
|
||||||
|
|
||||||
### Compare with Ridge
|
### Compare with Ridge
|
||||||
#### Find lambda
|
#### Find lambda
|
||||||
tune_results <- tune_ridge(bundled_data$train.X, bundled_data$train.y,
|
tune_results <- tune_ridge(bundled_data$train.X, bundled_data$train.y,
|
||||||
num_folds=10, 2^seq(-5,5,1), verbose=F)
|
num_folds=10, 2^seq(-5,5,1), verbose=F)
|
||||||
plot(log(tune_results$cv.table$hyp), tune_results$cv.table$means, type="l",
|
plot(log(tune_results$cv.table$hyp), tune_results$cv.table$means, type="l",
|
||||||
xlab="lambda", ylab="CV Mean Loss")
|
xlab="lambda", ylab="CV Mean Loss")
|
||||||
abline(v=tune_results$lam.min)
|
abline(v=tune_results$lam.min)
|
||||||
tune_results$lam.min
|
tune_results$lam.min
|
||||||
|
|
||||||
#### Use lam.min for Ridge Regression
|
#### Use lam.min for Ridge Regression
|
||||||
ridge_result <- ridge_betas(bundled_data$train.X, bundled_data$train.y,
|
ridge_result <- ridge_betas(bundled_data$train.X, bundled_data$train.y,
|
||||||
beta_init = NULL, lam=tune_results$lam.min, method="BFGS")
|
beta_init = NULL, lam=tune_results$lam.min, method="BFGS")
|
||||||
ridge.df <- data.frame(att=c("intercept", colnames(bundled_data$train.X)),
|
ridge.df <- data.frame(att=c("intercept", colnames(bundled_data$train.X)),
|
||||||
scores=ridge_result$betas,
|
scores=ridge_result$betas,
|
||||||
abs_scores=abs(ridge_result$betas))
|
abs_scores=abs(ridge_result$betas))
|
||||||
ridge.res <- dplyr::slice_max(ridge.df,order_by=abs_scores,n=20)
|
ridge.res <- dplyr::slice_max(ridge.df,order_by=abs_scores,n=20)
|
||||||
ridge.table <- as.data.table(ridge.res)
|
ridge.table <- as.data.table(ridge.res)
|
||||||
|
|
||||||
### Compare with Random Forest
|
### Compare with Random Forest
|
||||||
source("Schrick-Noah_Random-Forest.R")
|
source("Schrick-Noah_Random-Forest.R")
|
||||||
rf_result <- rf_comp(bundled_data$train)
|
rf_result <- rf_comp(bundled_data$train)
|
||||||
rf.df <- data.frame(att=c(colnames(bundled_data$train.X)),
|
rf.df <- data.frame(att=c(colnames(bundled_data$train.X)),
|
||||||
scores=rf_result$rf2_imp$rf_score)
|
scores=rf_result$rf2_imp$rf_score)
|
||||||
rf_res <- dplyr::slice_max(rf.df,order_by=scores, n=20)
|
rf_res <- dplyr::slice_max(rf.df,order_by=scores, n=20)
|
||||||
rf.table <- as.data.table(rf_res)
|
rf.table <- as.data.table(rf_res)
|
||||||
|
|
||||||
### Compare with glmnet
|
### Compare with glmnet
|
||||||
source("Schrick-Noah_glmnet.R")
|
source("Schrick-Noah_glmnet.R")
|
||||||
#### Alpha = 0
|
#### Alpha = 0
|
||||||
glm.res.0 <- glm_fcn(bundled_data$train.X, bundled_data$train.y, 0)
|
glm.res.0 <- glm_fcn(bundled_data$train.X, bundled_data$train.y, 0)
|
||||||
glm.df.0 <- data.frame(att=c("intercept", colnames(bundled_data$train.X)),
|
glm.df.0 <- data.frame(att=c("intercept", colnames(bundled_data$train.X)),
|
||||||
scores=glm.res.0$lambda.1se,
|
scores=glm.res.0$lambda.1se,
|
||||||
abs_scores=glm.res.0$abs_scores)
|
abs_scores=glm.res.0$abs_scores)
|
||||||
glm.df.0.res <- dplyr::slice_max(glm.df.0,order_by=abs_scores,n=20)
|
glm.df.0.res <- dplyr::slice_max(glm.df.0,order_by=abs_scores,n=20)
|
||||||
glm.0.table <- as.data.table(glm.df.0.res)
|
glm.0.table <- as.data.table(glm.df.0.res)
|
||||||
|
|
||||||
#### Alpha = 1
|
#### Alpha = 1
|
||||||
glm.res.1 <- glm_fcn(bundled_data$train.X, bundled_data$train.y, 1) # alpha=1
|
glm.res.1 <- glm_fcn(bundled_data$train.X, bundled_data$train.y, 1) # alpha=1
|
||||||
glm.df.1 <- data.frame(att=c("intercept", colnames(bundled_data$train.X)),
|
glm.df.1 <- data.frame(att=c("intercept", colnames(bundled_data$train.X)),
|
||||||
scores=glm.res.1$lambda.1se,
|
scores=glm.res.1$lambda.1se,
|
||||||
abs_scores=glm.res.1$abs_scores)
|
abs_scores=glm.res.1$abs_scores)
|
||||||
glm.df.1.res <- dplyr::slice_max(glm.df.1,order_by=abs_scores,n=20)
|
glm.df.1.res <- dplyr::slice_max(glm.df.1,order_by=abs_scores,n=20)
|
||||||
glm.1.table <- as.data.table(glm.df.1.res)
|
glm.1.table <- as.data.table(glm.df.1.res)
|
||||||
|
|
||||||
### Plot
|
### Plot
|
||||||
#### Convert names to indices
|
#### Convert names to indices
|
||||||
lasso.df$att <- match(lasso.df$att,colnames(bundled_data$train))
|
lasso.df$att <- match(lasso.df$att,colnames(bundled_data$train))
|
||||||
ridge.df$att <- match(ridge.df$att,colnames(bundled_data$train))
|
ridge.df$att <- match(ridge.df$att,colnames(bundled_data$train))
|
||||||
rf.df$att <- match(rf.df$att,colnames(bundled_data$train))
|
rf.df$att <- match(rf.df$att,colnames(bundled_data$train))
|
||||||
glm.df.0$att <- match(glm.df.0$att,colnames(bundled_data$train))
|
glm.df.0$att <- match(glm.df.0$att,colnames(bundled_data$train))
|
||||||
glm.df.1$att <- match(glm.df.1$att,colnames(bundled_data$train))
|
glm.df.1$att <- match(glm.df.1$att,colnames(bundled_data$train))
|
||||||
|
|
||||||
#### Scale
|
#### Scale
|
||||||
lasso.df$abs_scores <- scale(lasso.df$abs_scores)
|
lasso.df$abs_scores <- scale(lasso.df$abs_scores)
|
||||||
ridge.df$abs_scores <- scale(ridge.df$abs_scores)
|
ridge.df$abs_scores <- scale(ridge.df$abs_scores)
|
||||||
rf.df$scores <- scale(rf.df$scores)
|
rf.df$scores <- scale(rf.df$scores)
|
||||||
glm.df.0$abs_scores <- scale(glm.df.0$abs_scores)
|
glm.df.0$abs_scores <- scale(glm.df.0$abs_scores)
|
||||||
glm.df.1$abs_scores <- scale(glm.df.1$abs_scores)
|
glm.df.1$abs_scores <- scale(glm.df.1$abs_scores)
|
||||||
|
|
||||||
plot(x=lasso.df$att, y=lasso.df$abs_scores, type="l", xlab="Vars",
|
plot(x=lasso.df$att, y=lasso.df$abs_scores, type="l", xlab="Vars",
|
||||||
ylab="Coefficients (Abs Scores)", xaxt="n", col="blue", ylim=c(-1,3),
|
ylab="Coefficients (Abs Scores)", xaxt="n", col="blue", ylim=c(-1,3),
|
||||||
main="Scaled scores for simulated data feature selection")
|
main="Scaled scores for simulated data feature selection")
|
||||||
axis(1, at=1:101, labels=colnames(bundled_data$train), cex.axis=0.5)
|
axis(1, at=1:101, labels=colnames(bundled_data$train), cex.axis=0.5)
|
||||||
lines(x=ridge.df$att, y=ridge.df$abs_scores, col="red")
|
lines(x=ridge.df$att, y=ridge.df$abs_scores, col="red")
|
||||||
lines(x=rf.df$att, y=rf.df$scores, col="green")
|
lines(x=rf.df$att, y=rf.df$scores, col="green")
|
||||||
lines(x=glm.df.0$att, y=glm.df.0$abs_scores, col="bisque4")
|
lines(x=glm.df.0$att, y=glm.df.0$abs_scores, col="bisque4")
|
||||||
lines(x=glm.df.1$att, y=glm.df.1$abs_scores, col="purple")
|
lines(x=glm.df.1$att, y=glm.df.1$abs_scores, col="purple")
|
||||||
legend(x="topleft",
|
legend(x="topleft",
|
||||||
legend=c("LASSO", "Ridge", "Random Forest","glmnet (alpha=0)", "glmnet (alpha=1)"),
|
legend=c("LASSO", "Ridge", "Random Forest","glmnet (alpha=0)", "glmnet (alpha=1)"),
|
||||||
lty=c(1,1,1,1,1),
|
lty=c(1,1,1,1,1),
|
||||||
col=c("blue", "red", "green", "bisque4", "purple"),
|
col=c("blue", "red", "green", "bisque4", "purple"),
|
||||||
cex=1)
|
cex=1)
|
||||||
|
}
|
||||||
|
|
||||||
|
run_comparison(bundled_data)
|
||||||
## b. Repeat comparison using a graph with clusters
|
## b. Repeat comparison using a graph with clusters
|
||||||
if (!require("igraph")) install.packages("igraph")
|
source("Schrick-Noah_graphs.R")
|
||||||
library(igraph)
|
bundled_graph_data <- sim_graph_data()
|
||||||
if (!require("Matrix")) install.packages("Matrix")
|
run_comparison(bundled_graph_data)
|
||||||
library(Matrix) # bdiag
|
|
||||||
|
|
||||||
npc <-25 # nodes per cluster
|
|
||||||
n_clust <- 4 # 4 clusters with 25 nodes each
|
|
||||||
|
|
||||||
# no clusters
|
|
||||||
g0 <- erdos.renyi.game(npc*n_clust, 0.2)
|
|
||||||
plot(g0)
|
|
||||||
|
|
||||||
matlist = list()
|
|
||||||
for (i in 1:n_clust){
|
|
||||||
matlist[[i]] = get.adjacency(erdos.renyi.game(npc, 0.2))
|
|
||||||
}
|
|
||||||
|
|
||||||
# merge clusters into one matrix
|
|
||||||
mat_clust <- bdiag(matlist) # create block-diagonal matrix
|
|
||||||
|
|
||||||
## the following two things might not be necessary
|
|
||||||
|
|
||||||
# check for loner nodes, connected to nothing, and join them to something
|
|
||||||
k <- rowSums(mat_clust)
|
|
||||||
node_vector <- seq(1,npc*n_clust)
|
|
||||||
for (i in node_vector){
|
|
||||||
if (k[i]==0){ # if k=0, connect to something random
|
|
||||||
j <- sample(node_vector[-i],1)
|
|
||||||
mat_clust[i,j] <- 1
|
|
||||||
mat_clust[j,i] <- 1
|
|
||||||
}
|
|
||||||
}
|
|
||||||
|
|
||||||
node_colors <- c(rep("red",npc), rep("green",npc), rep("blue",npc), rep("orange",npc))
|
|
||||||
g1 <- graph_from_adjacency_matrix(mat_clust, mode="undirected", diag=F)
|
|
||||||
plot(g1, vertex.color=node_colors)
|
|
||||||
|
|
||||||
### Dataset with g1
|
|
||||||
dataset.graph <- npdro::createSimulation2(num.samples=num.samples,
|
|
||||||
num.variables=num.variables,
|
|
||||||
pct.imbalance=0.5,
|
|
||||||
pct.signals=0.2,
|
|
||||||
main.bias=0.5,
|
|
||||||
interaction.bias=1,
|
|
||||||
hi.cor=0.95,
|
|
||||||
lo.cor=0.2,
|
|
||||||
mix.type="main-interactionScalefree",
|
|
||||||
label="class",
|
|
||||||
sim.type="mixed",
|
|
||||||
pct.mixed=0.5,
|
|
||||||
pct.train=0.5,
|
|
||||||
pct.holdout=0.5,
|
|
||||||
pct.validation=0,
|
|
||||||
plot.graph=F,
|
|
||||||
graph.structure = g1,
|
|
||||||
verbose=T)
|
|
||||||
|
|
||||||
train.graph <- dataset.graph$train #150x101
|
|
||||||
test.graph <- dataset.graph$holdout
|
|
||||||
validation.graph <- dataset.graph$validation
|
|
||||||
dataset.graph$signal.names
|
|
||||||
colnames(train.graph)
|
|
||||||
|
|
||||||
# separate the class vector from the predictor data matrix
|
|
||||||
train.graph.X <- train.graph[, -which(colnames(train.graph) == "class")]
|
|
||||||
train.graph.y <- train.graph[, "class"]
|
|
||||||
train.graph.y.01 <- as.numeric(train.graph.y)-1
|
|
||||||
|
|
||||||
## c. Use npdro and igraph to create knn
|
## c. Use npdro and igraph to create knn
|
||||||
my.k <- 3 # larger k, fewer clusters
|
my.k <- 3 # larger k, fewer clusters
|
||||||
|
|||||||
43
Schrick-Noah_graphs.R
Normal file
43
Schrick-Noah_graphs.R
Normal file
@ -0,0 +1,43 @@
|
|||||||
|
source("Schrick-Noah_Simulated-Data.R")
|
||||||
|
|
||||||
|
if (!require("igraph")) install.packages("igraph")
|
||||||
|
library(igraph)
|
||||||
|
if (!require("Matrix")) install.packages("Matrix")
|
||||||
|
library(Matrix) # bdiag
|
||||||
|
|
||||||
|
sim_graph_data <- function(){
|
||||||
|
npc <-25 # nodes per cluster
|
||||||
|
n_clust <- 4 # 4 clusters with 25 nodes each
|
||||||
|
|
||||||
|
# no clusters
|
||||||
|
g0 <- erdos.renyi.game(npc*n_clust, 0.2)
|
||||||
|
plot(g0)
|
||||||
|
|
||||||
|
matlist = list()
|
||||||
|
for (i in 1:n_clust){
|
||||||
|
matlist[[i]] = get.adjacency(erdos.renyi.game(npc, 0.2))
|
||||||
|
}
|
||||||
|
|
||||||
|
# merge clusters into one matrix
|
||||||
|
mat_clust <- bdiag(matlist) # create block-diagonal matrix
|
||||||
|
|
||||||
|
## the following two things might not be necessary
|
||||||
|
# check for loner nodes, connected to nothing, and join them to something
|
||||||
|
k <- rowSums(mat_clust)
|
||||||
|
node_vector <- seq(1,npc*n_clust)
|
||||||
|
for (i in node_vector){
|
||||||
|
if (k[i]==0){ # if k=0, connect to something random
|
||||||
|
j <- sample(node_vector[-i],1)
|
||||||
|
mat_clust[i,j] <- 1
|
||||||
|
mat_clust[j,i] <- 1
|
||||||
|
}
|
||||||
|
}
|
||||||
|
|
||||||
|
node_colors <- c(rep("red",npc), rep("green",npc), rep("blue",npc), rep("orange",npc))
|
||||||
|
g1 <- graph_from_adjacency_matrix(mat_clust, mode="undirected", diag=F)
|
||||||
|
plot(g1, vertex.color=node_colors)
|
||||||
|
|
||||||
|
### Dataset with g1
|
||||||
|
bundled_graph_data <- create_data(graph.structure=g1)
|
||||||
|
return(bundled_graph_data)
|
||||||
|
}
|
||||||
Loading…
x
Reference in New Issue
Block a user