Revision e2d673e4
Added by Benoit Parmentier about 12 years ago
- ID e2d673e42b81d3fc56e248b6009ad94a9b5306b4
| climate/research/oregon/interpolation/Linear_reg.R | ||
|---|---|---|
| 1 |
#This script interpolates station value for the Oregon case study. This program loads the station data from a csv file |
|
| 1 |
#This script interpolates station value for the Oregon case study. This program loads the station data from a csv file
|
|
| 2 | 2 |
#and perform two types of regression: multiple linear model and general additive model (GAM). Note that this program: |
| 3 |
#1)assumes that the csv file is in the current working |
|
| 4 |
#2)extract relevant variables from raster images before performing the regressions. |
|
| 3 |
#1)assumes that the csv file is in the current working
|
|
| 4 |
#2)extract relevant variables from raster images before performing the regressions.
|
|
| 5 | 5 |
#The user must provide the list of raster images in a textile. |
| 6 |
#Script created by Benoit Parmentier on February 15, 2012.
|
|
| 6 |
#Script created by Benoit Parmentier on February 28, 2012.
|
|
| 7 | 7 |
|
| 8 |
###Loading r library and packages |
|
| 9 |
library(raster) # loading the raster package |
|
| 8 |
###Loading r library and packages # loading the raster package |
|
| 10 | 9 |
library(gtools) # loading ... |
| 11 |
library(sp) |
|
| 12 | 10 |
library(mgcv) |
| 13 | 11 |
|
| 14 | 12 |
###Parameters and arguments |
| 15 | 13 |
infile1<-"ghcn_or_b_02122012_OR83M.csv" |
| 16 |
inlistf<-"list_files.txt" |
|
| 17 | 14 |
path<-"C:/Data/Benoit/NCEAS/window_Oregon_data" |
| 18 |
#path<-getwd() |
|
| 19 |
#date<-"20101507" #Date selected for the regression |
|
| 20 |
#prop<-0.3 #Proportion of testing retained for validation |
|
| 21 |
#list_var<-list.files() # displaying the files in the current folder which should contain only the relevant raster images. |
|
| 22 |
#dump("list_var", file="list_var.txt") # Saving the values of list_var in a ascii text file.
|
|
| 23 |
#inlistvar<-"ghcn_var_02122012.txt" |
|
| 15 |
setwd(path) |
|
| 16 |
#date<-"20101507" #Date selected for the regression |
|
| 17 |
#prop<-0.3 #Proportion of testing retained for validation |
|
| 24 | 18 |
|
| 25 | 19 |
#######START OF THE SCRIPT ############# |
| 26 | 20 |
|
| 27 | 21 |
###Reading the station data |
| 28 |
ghcn<-read.csv(paste(path,"/",infile1, sep=""), header=TRUE) #The "paste" function concatenates the path and file name in common string. |
|
| 22 |
ghcn<-read.csv(paste(path,"/",infile1, sep=""), header=TRUE) #The "paste" function concatenates the path and file name in common string.
|
|
| 29 | 23 |
#ghcn<-read.csv(infile1) #Use read.table if the input file is space delimited or read.table(infile1, headername=TRUE, sep=',') |
| 30 | 24 |
names(ghcn) #Checking that the columns are correctly labelled. |
| 31 | 25 |
|
| 32 |
###Extracting the variables values from the raster files |
|
| 33 |
coords<- ghcn[,c('x_OR83M','y_OR83M')] #Selecting all rows for the x and y columns and packaging the output in a data frame.
|
|
| 34 |
lines <-readLines(paste(path,"/",inlistf, sep="") |
|
| 35 |
inlistvar<-paste(path,"/",lines,sep="") |
|
| 36 |
|
|
| 37 |
#s_raster<- stack(inlistvar) #Creating a stack of raster images from the list of variables. |
|
| 38 |
#stat_val<- extract(s_raster, coords) #Extracting values from the raster stack for every point location in coords data frame. |
|
| 39 |
#create a shape file and data_frame with names?? |
|
| 40 |
|
|
| 41 | 26 |
###Creating a validation dataset by creating training and testing datasets (%30) |
| 42 | 27 |
#ghcn1507 <-subset(ghcn,ghcn$date_== date) |
| 43 | 28 |
ghcn1507 <-subset(ghcn,ghcn$date_=="20100715") |
| ... | ... | |
| 69 | 54 |
|
| 70 | 55 |
AIC (lm_ANUSPLIN1,GAM_ANUSPLIN1) #list the AIC and for the results |
| 71 | 56 |
AIC (lm_PRISM1,GAM_PRISM1) #list the AIC and for the results |
| 72 |
anova(lm_ANUSPLIN1, GAM_ANUSPLIN1,test="F") #compare the different models in terms of F; a reference model should be set for comparison. |
|
| 57 |
|
|
| 58 |
GCVA1<-GAM_ANUSPLIN1$gcv.ubre |
|
| 59 |
GCVP1<-GAM_PRISM1$gcv.ubre |
|
| 60 |
|
|
| 61 |
DevianceA1<-GAM_ANUSPLIN1$deviance |
|
| 62 |
DevianceP1<-GAM_PRISM1$deviance |
|
| 63 |
|
|
| 64 |
anova(lm_ANUSPLIN1, GAM_ANUSPLIN1,test="F") #compare the different models in terms of F; a reference model should be set for comparison. |
|
| 73 | 65 |
anova(lm_PRISM1, GAM_PRISM1,test="F") |
| 74 | 66 |
|
| 75 |
#GAM plot of partial residuals |
|
| 67 |
summary(lm_ANUSPLIN1)$r.squared |
|
| 68 |
summary(GAM_ANUSPLIN1)$r.squared |
|
| 69 |
summary(lm_PRISM1)$r.squared |
|
| 70 |
summary(GAM_PRISM1)$r.squared |
|
| 71 |
|
|
| 72 |
############Diagnostic GAM plots############# |
|
| 73 |
win.graph() |
|
| 74 |
gam.check(GAM_ANUSPLIN1) |
|
| 75 |
savePlot("GAM_ANUSPLIN1_diagnostic1.emf", type="emf")
|
|
| 76 |
win.graph() |
|
| 76 | 77 |
gam.check(GAM_PRISM1) #This will produce basic plots of residuals |
| 78 |
savePlot("GAM_PRISM_diagnostic1.emf", type="emf")
|
|
| 77 | 79 |
gam.check(GAM_ANUSPLIN1) |
| 78 |
|
|
| 79 |
#Prediction checking the results using the testing data |
|
| 80 |
win.graph() |
|
| 81 |
vis.gam(GAM_ANUSPLIN1) |
|
| 82 |
savePlot("GAM_ANUSPLIN1_prediction.emf", type="emf")
|
|
| 83 |
win.graph() |
|
| 84 |
vis.gam(GAM_PRISM1) |
|
| 85 |
savePlot("GAM_PRISM1_prediction.emf", type="emf")
|
|
| 86 |
win.graph() |
|
| 87 |
vis.gam(GAM_ANUSPLIN1, view=c("lat","ELEV_SRTM"))
|
|
| 88 |
#vis.gam(GAM_ANUSPLIN1, view=c("lat","ELEV_SRTM", theta=100,phi=200))
|
|
| 89 |
savePlot("GAM_ANUSPLIN1_prediction2.emf", type="emf")
|
|
| 90 |
|
|
| 91 |
####VALIDATION: Prediction checking the results using the testing data######## |
|
| 80 | 92 |
y_plANUSPLIN1<- predict(lm_ANUSPLIN1, newdata=ghcn1507_v, se.fit = TRUE) |
| 81 |
y_pgANUSPLIN1<- predict(GAM_ANUSPLIN1, newdata=ghcn1507_v, se.fit = TRUE) |
|
| 93 |
y_pgANUSPLIN1<- predict(GAM_ANUSPLIN1, newdata=ghcn1507_v, se.fit = TRUE) #Using the coeff to predict new values.
|
|
| 82 | 94 |
y_plPRISM1<- predict(lm_PRISM1, newdata=ghcn1507_v, se.fit = TRUE) |
| 83 |
y_pgPRISM1<- predict(GAM_PRISM1, newdata=ghcn1507_v, se.fit = TRUE) |
|
| 84 |
|
|
| 95 |
y_pgPRISM1<- predict(GAM_PRISM1, newdata=ghcn1507_v, se.fit = TRUE)
|
|
| 96 |
|
|
| 85 | 97 |
res_yplA1<- ghcn1507_v$tmax - y_plANUSPLIN1$fit |
| 86 | 98 |
res_ypgA1<- ghcn1507_v$tmax - y_pgANUSPLIN1$fit |
| 87 | 99 |
res_yplP1<- ghcn1507_v$tmax - y_plPRISM1$fit #Residuals for lm model that resembles the PRISM interpolation |
| 88 |
res_ypgP1<- ghcn1507_v$tmax - y_pgPRISM1$fit #Residuals for GAM model that resembles the PRISM interpolation |
|
| 89 |
|
|
| 90 |
RMSE_yplA1 <- sqrt(sum(res_yplA1^2)) |
|
| 91 |
#RMSE_ypgA1 <- sqrt(sum(res_ypgA1^2)) |
|
| 92 |
RMSE_ypgA1 <- sqrt(sum(res_ypgA1^2)) |
|
| 93 |
RMSE_yplP1 <- sqrt(sum(res_yplP1^2)) |
|
| 94 |
RMSE_ypgP1 <- sqrt(sum(res_ypgP1^2)) |
|
| 95 |
|
|
| 96 |
#Printing the RMSE values for the different models |
|
| 97 |
RMSE_yplA1 |
|
| 98 |
RMSE_ypgA1 |
|
| 99 |
RMSE_yplP1 |
|
| 100 |
res_ypgP1<- ghcn1507_v$tmax - y_pgPRISM1$fit #Residuals for GAM model that resembles the PRISM interpolation |
|
| 101 |
|
|
| 102 |
#results_val <-c(res_yplA1,res_ypgA1,res_yplP1,res_ypgP1) |
|
| 103 |
results_val<-data.frame(res_yplA1=res_yplA1,res_ypgA1=res_ypgA1,res_yplP1=res_yplP1,res_ypgP1=res_ypgP1) |
|
| 104 |
nv<- nrow(ghcn1507_v) |
|
| 105 |
|
|
| 106 |
RMSE_yplA1 <- sqrt(sum(res_yplA1^2)*1/nv) |
|
| 107 |
#RMSE_ypgA1 <- sqrt(sum(res_ypgA1^2)/nv) |
|
| 108 |
RMSE_ypgA1 <- sqrt(sum(res_ypgA1^2)/nv) |
|
| 109 |
RMSE_yplP1 <- sqrt(sum(res_yplP1^2)/nv) |
|
| 110 |
RMSE_ypgP1 <- sqrt(sum(res_ypgP1^2)/nv) |
|
| 111 |
|
|
| 112 |
#Printing the RMSE values for the different models |
|
| 113 |
RMSE_yplA1 |
|
| 114 |
RMSE_ypgA1 |
|
| 115 |
RMSE_yplP1 |
|
| 100 | 116 |
RMSE_ypgP1 |
| 101 | 117 |
|
| 102 |
RMSE_all<-c(RMSE_yplA1,RMSE_ypgA1,RMSE_yplP1,RMSE_ypgP1) |
|
| 103 | 118 |
mod_name<-c("yplA1","ypgA1","yplP1","ypgP1")
|
| 104 | 119 |
mod_type<-c("lm_ANUSPLIN1","GAM_ANUSPLIN1","lm_PRISM1","GAM_PRISM1")
|
| 120 |
|
|
| 121 |
RMSE_all<-c(RMSE_yplA1,RMSE_ypgA1,RMSE_yplP1,RMSE_ypgP1) |
|
| 105 | 122 |
AIC_all<-AIC(lm_ANUSPLIN1,GAM_ANUSPLIN1,lm_PRISM1,GAM_PRISM1) |
| 106 |
results<-data.frame(model=mod_name,RMSE=RMSE_all,df=AIC_all$df,AIC=AIC_all$AIC) |
|
| 123 |
GCV_all<-c(0,GCVA1,0,GCVP1) #This places the GCV values for each model in a vector |
|
| 124 |
Deviance_all<-c(0,DevianceA1,0,DevianceP1) |
|
| 125 |
|
|
| 126 |
#results<-data.frame(model=mod_name,RMSE=RMSE_all,df=AIC_all$df,AIC=AIC_all$AIC) |
|
| 127 |
#results_val<-data.frame(model=mod_name,RMSE=RMSE_all,df=AIC_all$df,AIC=AIC_all$AIC) |
|
| 128 |
results<-data.frame(model=mod_name,RMSE=RMSE_all,df=AIC_all$df,AIC=AIC_all$AIC, GCV=GCV_all,Deviance=Deviance_all) |
|
| 129 |
|
|
| 130 |
#Add deviance |
|
| 107 | 131 |
|
| 132 |
barplot(results$RMSE,main="RMSE for the models",names.arg=c("yplA1","ypgA1","yplP1","ypgP1"),ylab="Temp (deg. C)")
|
|
| 133 |
barplot(results$AIC,main="AIC for the models",names.arg=c("yplA1","ypgA1","yplP1","ypgP1"),ylab="AIC")
|
|
| 108 | 134 |
#dump("results", file= paste(path,"/","results_reg1507.txt",sep=""))
|
| 109 |
write.csv(results, file= paste(path,"/","results_reg1507.txt",sep="")) |
|
| 135 |
write.csv(results, file= paste(path,"/","results_reg1507_Assessment.txt",sep="")) |
|
| 136 |
#write.csv(results_val, file= paste(path,"/","results_reg1507_val.txt",sep="")) |
|
| 110 | 137 |
|
| 111 |
##End of script |
|
| 138 |
|
|
| 139 |
library(lattice) |
|
| 140 |
g <- expand.grid(x = ghcn1507_v$lon, y = ghcn1507_v$lat) |
|
| 141 |
g$z <- y_plANUSPLIN1$fit |
|
| 142 |
wireframe(z ~ x * y, data = g) |
|
| 112 | 143 |
|
| 144 |
#####End of script########## |
|
| 145 |
############################### |
|
| 113 | 146 |
|
| 114 | 147 |
|
| 115 | 148 |
|
| 116 |
|
|
| 149 |
|
|
Also available in: Unified diff
Corrected errors in RMSE calculation and added plots to view results