-
Notifications
You must be signed in to change notification settings - Fork 1
/
metrics.R
85 lines (68 loc) · 2.69 KB
/
metrics.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
# MSc Project 2020: Comparison of Spatial Gaussian Copula and Random Forests in Spatial Prediction
# Sun, N
#
# Code for calculating prediction metrics
rmspe <- function(predicted_values, observed_values) {
#' Computes root mean squared prediction error
#'
#' @param predicted_values vector of predicted values
#' @param observed_values vector of observed values
#'
#' @return RMSPE single value
if(length(predicted_values) != length(observed_values)) {
stop("Vectors are of unequal length.")
}
sq.diffs <- (predicted_values - observed_values)^2
return(sqrt(mean(sq.diffs)))
}
mspe <- function(predicted_values, observed_values) {
#' Computes sum of squared residuals
#'
#' @param predicted_values vector of predicted values
#' @param observed_values vector of observed values
#'
#' @return RMSPE single value
if(length(predicted_values) != length(observed_values)) {
stop("Vectors are of unequal length.")
}
sum.sq.diffs <- sum((predicted_values - observed_values)^2)
return((sum.sq.diffs))
}
srb <- function(predicted_values, observed_values) {
#' Computes Signed Relative Bias for predictions
#' using formula from ver Hoef (2013)
#'
#' @param predicted_values vector of predicted values
#' @param observed_values vector of observed values
#'
#' @return signed relative bias
if(length(predicted_values) != length(observed_values)) {
stop("Vectors are of unequal length.")
}
tau <- mean((predicted_values - observed_values))
srb <- sign(tau)*sqrt(tau^2/(rmspe(predicted_values, observed_values)^2 - tau^2))
return(srb)
}
pic90 <- function(predicted_values, observed_values) {
#' Computes 90% prediction interval coverage
#' using formula from ver Hoef (2013).
#'
#' The value produced here should be close to 90%.
#'
#' @param predicted_values vector of predicted values
#' @param observed_values vector of observed values
#'
#' @return signed relative bias
se <- sd(predicted_values)
upper.bound <- predicted_values + 1.645*se
lower.bound <- predicted_values - 1.645*se
coverage <- as.integer((observed_values >= lower.bound) & (observed_values <= upper.bound))
return(mean(coverage))
}
make_plot <- function(data) {
ggplot(data, aes(x=x/1000,y=y/1000,color=resp)) +
geom_point(size=2) +
scale_color_gradient(low = "white", high = "red") +
scale_x_continuous(breaks = round(seq(-2150, -2000, by = 50),2)) +
theme(axis.text.x=element_text(angle=90,hjust=1))
}