-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathLungQtrlyClass.R
131 lines (115 loc) · 4.48 KB
/
LungQtrlyClass.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
library(raster)
library(rgdal)
library(sp)
library(randomForest)
library(RStoolbox)
library(e1071)
library(reshape2)
library(cluster)
library(terra)
library(rpart)
library(caTools)
library(fields)
library(dplyr)
library(purrr)
library(magick)
library(caret)
setwd("")
set.seed(1)
#load the 2019 geomedians
LungQ1 <- stack(".tif")
LungQ2 <- stack(".tif")
LungQ3 <- stack(".tif")
LungQ4 <- stack(".tif")
summary(LungQ1)
s2bands <- list("green", "red", "blue", "nir", "swir_1", "swir_2", "SMAD", "NDVI", "NDWI", "BSI")
LungQ1 <- setNames(LungQ1, s2bands)
LungQ1plot <- plotRGB(LungQ1, r=2, g=1, b=3, stretch="hist")
summary(LungQ1)
LungQ2 <- setNames(LungQ2, s2bands)
LungQ3 <- setNames(LungQ3, s2bands)
LungQ4 <- setNames(LungQ4, s2bands)
#StackQtrs
LungQtrs <- stack(LungQ1,LungQ2,LungQ3,LungQ4)
#load training polygons
LungTrain <- readOGR(".shp")
summary(LungTrain)
plot(LungTrain)
LungTrain$Class <- as.factor(LungTrain$id)
classes <- c("Irrigated","Other")
classesdf <- data.frame(classnum = c(1,2), classnames=classes)
levels(LungTrain$Class) <- c("Irrigated","Other")
#' Divide data into training and validation
Lung_smp_size <- floor(0.8*nrow(LungTrain))
Lung_train_ind <- sample(seq_len(nrow(LungTrain)), size=Lung_smp_size)
LungTrain_tr <- LungTrain[Lung_train_ind,]
LungTrain_val <- LungTrain[-Lung_train_ind,]
#' RStoolbox superclass
#superclassLung <- superClass(LungQtrs, LungTrain, valData = NULL, responseCol = "id",
# nSamples = 100000, polygonBasedCV = TRUE, trainPartition = 0.8,
#model = "rf", tuneLength = 2, kfold = 5, minDist = 1,
# mode = "classification", predict = FALSE, predType = "raw",
# filename = NULL, verbose=TRUE, overwrite = TRUE)
#superclassLung
#superpredictLung <- predict(superclassLung, LungQtrs)
#plot(superpredictLung)
#superclassimp <- varImp(superclassLung$model)
#plot(superclassimp, type = 0)
#varImpPlot(superclassLung$model, type=1)
#' Generate point samples on training data
pt_Lungtrain <- spsample(LungTrain_tr, 80000, type='random')
pt_Lungtrain$Class <- over(pt_Lungtrain, LungTrain_tr)$Class
pt_Lungtrain <- vect(pt_Lungtrain)
#' Extract spectral vals on training data
xy_train <- as.matrix(geom(pt_Lungtrain)[,c('x','y')])
df_Lungtrain <- extract(LungQtrs, xy_train)
head(df_Lungtrain)
data_Lungtrain <- data.frame(Class=pt_Lungtrain$Class, df_Lungtrain)
summary(data_Lungtrain)
data_Lungtrain <- na.omit(data_Lungtrain)
#' Generate point samples on validation data
pt_Lungval <- spsample(LungTrain_val, 20000, type='random')
pt_Lungval$Class <- over(pt_Lungval, LungTrain_val)$Class
pt_Lungval <- vect(pt_Lungval)
#' Extract spectral vals on validation data
xy_val <- as.matrix(geom(pt_Lungval)[,c('x','y')])
df_Lungval <- extract(LungQtrs, xy_val)
head(df_Lungval)
data_Lungval <- data.frame(Class=pt_Lungval$Class, df_Lungval)
summary(data_Lungval)
data_Lungval <- na.omit(data_Lungval)
#' Create training and validation data
LungTraining = data_Lungtrain
LungValid = data_Lungval
#' RF classifier
Lung_Random_Forest_Classification <- randomForest(x=LungTraining[-1], y=as.factor(LungTraining$Class),
ntree=500, importance=TRUE)
plot(Lung_Random_Forest_Classification)
Lung_Random_Forest_Classification
#' Variable importance
LungQtrVIP <- varImpPlot(Lung_Random_Forest_Classification, type=1)
LungQtrImp <- as.data.frame(LungQtrVIP)
LungQtrImp$varnames <- rownames(LungQtrImp)
rownames(LungQtrImp) <- NULL
LungQtrImp
LungQtrVIPgg <- ggplot(LungQtrImp, aes(x=reorder(varnames, MeanDecreaseAccuracy), y=MeanDecreaseAccuracy))+
geom_point()+geom_segment(aes(x=varnames,xend=varnames,y=0,yend=MeanDecreaseAccuracy))+
xlab("Variable")+ylab("Mean Decrease Accuracy")+coord_flip()
LungQtrVIPgg
#' Prediction stats and confusion matrix
LungPrediction <- predict(Lung_Random_Forest_Classification, newdata=LungValid)
LungconMat <- confusionMatrix(LungPrediction, as.factor(LungValid$Class))
LungconMat
#' Look at the map
LungMap <- predict(LungQtrs, Lung_Random_Forest_Classification, filename="RF_Lung.img", type="response",
index=1, na.rm=TRUE, progress="window", overwrite=TRUE)
plot(LungMap)
LungMap <- ratify(LungMap)
LungMap
levels(LungMap$RF_Lung) <- c("Irrigated","Other")
library(rasterVis)
LungMap <- ratify(LungMap)
rat <- levels(LungMap)[[1]]
rat$legend <- classesdf$classnames
levels(LungMap) <- rat
LungLevelMap <- levelplot(LungMap)