demo/smoothing_params.R

In yinscapital/retina-analysis-reference: Tools for plotting and analyzing vertebrate retinas.

RES_variable <- 512

# pdf(file="smoothing_combos.pdf", width=8, height=5.5, family="Palatino")
	#choose combinations of the polynomial degree (m) and lambda
	poly <-c(2, 3, 4, 5)
	lambda <- c(1e-2,1e-3,1e-4,1e-5)
	# Generate error vectors for each combination
	# Save each plot to a PDF in the working directory
	ERR_dat<-polynomial_vs_lambda(Ntae_381, spatial_res= RES_variable, polynomial_m_vec=poly, lambda_vec=lambda)

sampling_location_number = length(Ntae_381$fit_data1$y)

ERR_dat <- ERR_dat[-1] #take off the NA placeholder column


X_median	<- retina:::reorder_columns(ERR_dat, median)
X_min 		<- retina:::reorder_columns(ERR_dat, min)
X_max  		<- retina:::reorder_columns(ERR_dat, max)
X_mean  	<- retina:::reorder_columns(ERR_dat, mean)
X_sd  		<- retina:::reorder_columns(ERR_dat, sd)
X_range_len <- retina:::reorder_columns(ERR_dat, retina:::range_len)



the_types <- c( 'X_median',
				'X_min',
				'X_max',
				'X_mean',
				'X_sd',
				'X_range_len')

for (e in the_types) {
	err_obj <- eval(parse(text=e))
	boxplot(err_obj, horizontal=TRUE, las=1, col='lightgray', cex=.2, pch=20, sub=paste("n = ", sampling_location_number), cex.axis=0.4, main=paste0("Ordered by ", e))
}

boxplot(X_sd, horizontal=TRUE, las=1, col='lightgray', asp=1, cex=.2, pch=20, sub=paste("n = ", sampling_location_number), cex.axis=0.4, main=paste0("Ordered by sd"))

barplot(apply(X_sd, 2 ,sd), horiz=TRUE, las=1)