anofox-regression 0.5.13

A robust statistics library for regression analysis
Documentation 
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
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261

#!/usr/bin/env Rscript
# Generate validation data for PLS (Partial Least Squares) regression
# This script creates test cases with known outputs from R's pls package

# Install pls package if not available
if (!require("pls", quietly = TRUE)) {
  install.packages("pls", repos = "https://cloud.r-project.org")
  library(pls)
}

set.seed(42)

cat("// =============================================================================\n")
cat("// PLS Regression Validation Data Generated from R\n")
cat(sprintf("// Generated on: %s\n", format(Sys.time())))
cat(sprintf("// R version: %s\n", R.version.string))
cat(sprintf("// pls package version: %s\n", packageVersion("pls")))
cat("// =============================================================================\n\n")

# -----------------------------------------------------------------------------
# Test 1: Simple PLS with 2 components
# -----------------------------------------------------------------------------
cat("// -----------------------------------------------------------------------------\n")
cat("// Test 1: Simple PLS with 2 components (SIMPLS)\n")
cat("// -----------------------------------------------------------------------------\n")

n <- 30
p <- 5
ncomp <- 2

# Generate correlated predictors
X1 <- matrix(rnorm(n * p), n, p)
# Add correlation structure
X1[,2] <- X1[,1] * 0.7 + rnorm(n, sd = 0.3)
X1[,4] <- X1[,3] * 0.8 + rnorm(n, sd = 0.2)

# True relationship
beta_true <- c(1.5, 0.8, -1.2, 0.4, 0.6)
y1 <- X1 %*% beta_true + rnorm(n, sd = 0.5)

# Create data frame for pls
df1 <- data.frame(y = y1, X = I(X1))

# Fit PLS with SIMPLS algorithm
fit1 <- plsr(y ~ X, data = df1, ncomp = ncomp, method = "simpls",
             scale = FALSE, center = TRUE)

# Get coefficients for ncomp components
coefs1 <- as.vector(coef(fit1, ncomp = ncomp))
intercept1 <- mean(y1) - sum(colMeans(X1) * coefs1)

# Get predictions on training data
fitted1 <- as.vector(predict(fit1, ncomp = ncomp))

# Compute R² manually (same way as our implementation)
y_mean1 <- mean(y1)
tss1 <- sum((y1 - y_mean1)^2)
rss1 <- sum((y1 - fitted1)^2)
r2_1 <- 1 - rss1 / tss1

cat(sprintf("// R Code: plsr(y ~ X, ncomp = %d, method = \"simpls\", scale = FALSE)\n", ncomp))
cat(sprintf("const N_PLS1: usize = %d;\n", n))
cat(sprintf("const P_PLS1: usize = %d;\n", p))
cat(sprintf("const NCOMP_PLS1: usize = %d;\n", ncomp))
cat("#[rustfmt::skip]\n")
cat(sprintf("const X_PLS1: [f64; %d] = [\n    ", n * p))
cat(paste(sprintf("%.10f", as.vector(X1)), collapse = ", "))
cat("\n];\n")
cat("#[rustfmt::skip]\n")
cat(sprintf("const Y_PLS1: [f64; %d] = [\n    ", n))
cat(paste(sprintf("%.10f", as.vector(y1)), collapse = ", "))
cat("\n];\n")
cat(sprintf("const EXPECTED_INTERCEPT_PLS1: f64 = %.10f;\n", intercept1))
cat("#[rustfmt::skip]\n")
cat(sprintf("const EXPECTED_COEFS_PLS1: [f64; %d] = [\n    ", p))
cat(paste(sprintf("%.10f", coefs1), collapse = ", "))
cat("\n];\n")
cat(sprintf("const EXPECTED_R2_PLS1: f64 = %.10f;\n", r2_1))
cat("\n")

# -----------------------------------------------------------------------------
# Test 2: PLS with more components
# -----------------------------------------------------------------------------
cat("// -----------------------------------------------------------------------------\n")
cat("// Test 2: PLS with 4 components (SIMPLS)\n")
cat("// -----------------------------------------------------------------------------\n")

n <- 50
p <- 8
ncomp <- 4

# Generate predictors with more complex structure
X2 <- matrix(rnorm(n * p), n, p)
# Add some correlation
for (i in 2:p) {
  X2[,i] <- X2[,i] + 0.3 * X2[,1]
}

# True relationship (sparse)
beta_true2 <- c(2.0, -1.5, 0.0, 1.0, 0.0, -0.5, 0.3, 0.0)
y2 <- X2 %*% beta_true2 + rnorm(n, sd = 1.0)

# Create data frame for pls
df2 <- data.frame(y = y2, X = I(X2))

# Fit PLS with SIMPLS algorithm
fit2 <- plsr(y ~ X, data = df2, ncomp = ncomp, method = "simpls",
             scale = FALSE, center = TRUE)

# Get coefficients for ncomp components
coefs2 <- as.vector(coef(fit2, ncomp = ncomp))
intercept2 <- mean(y2) - sum(colMeans(X2) * coefs2)

# Get predictions on training data
fitted2 <- as.vector(predict(fit2, ncomp = ncomp))

# Compute R²
y_mean2 <- mean(y2)
tss2 <- sum((y2 - y_mean2)^2)
rss2 <- sum((y2 - fitted2)^2)
r2_2 <- 1 - rss2 / tss2

cat(sprintf("// R Code: plsr(y ~ X, ncomp = %d, method = \"simpls\", scale = FALSE)\n", ncomp))
cat(sprintf("const N_PLS2: usize = %d;\n", n))
cat(sprintf("const P_PLS2: usize = %d;\n", p))
cat(sprintf("const NCOMP_PLS2: usize = %d;\n", ncomp))
cat("#[rustfmt::skip]\n")
cat(sprintf("const X_PLS2: [f64; %d] = [\n    ", n * p))
cat(paste(sprintf("%.10f", as.vector(X2)), collapse = ", "))
cat("\n];\n")
cat("#[rustfmt::skip]\n")
cat(sprintf("const Y_PLS2: [f64; %d] = [\n    ", n))
cat(paste(sprintf("%.10f", as.vector(y2)), collapse = ", "))
cat("\n];\n")
cat(sprintf("const EXPECTED_INTERCEPT_PLS2: f64 = %.10f;\n", intercept2))
cat("#[rustfmt::skip]\n")
cat(sprintf("const EXPECTED_COEFS_PLS2: [f64; %d] = [\n    ", p))
cat(paste(sprintf("%.10f", coefs2), collapse = ", "))
cat("\n];\n")
cat(sprintf("const EXPECTED_R2_PLS2: f64 = %.10f;\n", r2_2))
cat("\n")

# -----------------------------------------------------------------------------
# Test 3: PLS with scaling
# -----------------------------------------------------------------------------
cat("// -----------------------------------------------------------------------------\n")
cat("// Test 3: PLS with scaling (SIMPLS)\n")
cat("// -----------------------------------------------------------------------------\n")

n <- 40
p <- 6
ncomp <- 3

# Generate predictors with different scales
X3 <- matrix(rnorm(n * p), n, p)
X3[,1] <- X3[,1] * 100    # Large scale
X3[,2] <- X3[,2] * 0.01   # Small scale
X3[,3] <- X3[,3] * 1000   # Very large scale

# True relationship
beta_true3 <- c(0.02, 50.0, 0.001, 1.0, -0.5, 0.3)
y3 <- X3 %*% beta_true3 + rnorm(n, sd = 2.0)

# Create data frame for pls
df3 <- data.frame(y = y3, X = I(X3))

# Fit PLS with SIMPLS algorithm WITH scaling
fit3 <- plsr(y ~ X, data = df3, ncomp = ncomp, method = "simpls",
             scale = TRUE, center = TRUE)

# Get coefficients for ncomp components (on original scale)
coefs3 <- as.vector(coef(fit3, ncomp = ncomp))
intercept3 <- mean(y3) - sum(colMeans(X3) * coefs3)

# Get predictions on training data
fitted3 <- as.vector(predict(fit3, ncomp = ncomp))

# Compute R²
y_mean3 <- mean(y3)
tss3 <- sum((y3 - y_mean3)^2)
rss3 <- sum((y3 - fitted3)^2)
r2_3 <- 1 - rss3 / tss3

cat(sprintf("// R Code: plsr(y ~ X, ncomp = %d, method = \"simpls\", scale = TRUE)\n", ncomp))
cat(sprintf("const N_PLS3: usize = %d;\n", n))
cat(sprintf("const P_PLS3: usize = %d;\n", p))
cat(sprintf("const NCOMP_PLS3: usize = %d;\n", ncomp))
cat("#[rustfmt::skip]\n")
cat(sprintf("const X_PLS3: [f64; %d] = [\n    ", n * p))
cat(paste(sprintf("%.10f", as.vector(X3)), collapse = ", "))
cat("\n];\n")
cat("#[rustfmt::skip]\n")
cat(sprintf("const Y_PLS3: [f64; %d] = [\n    ", n))
cat(paste(sprintf("%.10f", as.vector(y3)), collapse = ", "))
cat("\n];\n")
cat(sprintf("const EXPECTED_INTERCEPT_PLS3: f64 = %.10f;\n", intercept3))
cat("#[rustfmt::skip]\n")
cat(sprintf("const EXPECTED_COEFS_PLS3: [f64; %d] = [\n    ", p))
cat(paste(sprintf("%.10f", coefs3), collapse = ", "))
cat("\n];\n")
cat(sprintf("const EXPECTED_R2_PLS3: f64 = %.10f;\n", r2_3))
cat("\n")

# -----------------------------------------------------------------------------
# Test 4: Small dataset with single component
# -----------------------------------------------------------------------------
cat("// -----------------------------------------------------------------------------\n")
cat("// Test 4: Small dataset with 1 component (SIMPLS)\n")
cat("// -----------------------------------------------------------------------------\n")

n <- 15
p <- 3
ncomp <- 1

X4 <- matrix(rnorm(n * p), n, p)
beta_true4 <- c(2.0, 1.0, 0.5)
y4 <- X4 %*% beta_true4 + rnorm(n, sd = 0.3)

df4 <- data.frame(y = y4, X = I(X4))
fit4 <- plsr(y ~ X, data = df4, ncomp = ncomp, method = "simpls",
             scale = FALSE, center = TRUE)

coefs4 <- as.vector(coef(fit4, ncomp = ncomp))
intercept4 <- mean(y4) - sum(colMeans(X4) * coefs4)
fitted4 <- as.vector(predict(fit4, ncomp = ncomp))

y_mean4 <- mean(y4)
tss4 <- sum((y4 - y_mean4)^2)
rss4 <- sum((y4 - fitted4)^2)
r2_4 <- 1 - rss4 / tss4

cat(sprintf("// R Code: plsr(y ~ X, ncomp = %d, method = \"simpls\", scale = FALSE)\n", ncomp))
cat(sprintf("const N_PLS4: usize = %d;\n", n))
cat(sprintf("const P_PLS4: usize = %d;\n", p))
cat(sprintf("const NCOMP_PLS4: usize = %d;\n", ncomp))
cat("#[rustfmt::skip]\n")
cat(sprintf("const X_PLS4: [f64; %d] = [\n    ", n * p))
cat(paste(sprintf("%.10f", as.vector(X4)), collapse = ", "))
cat("\n];\n")
cat("#[rustfmt::skip]\n")
cat(sprintf("const Y_PLS4: [f64; %d] = [\n    ", n))
cat(paste(sprintf("%.10f", as.vector(y4)), collapse = ", "))
cat("\n];\n")
cat(sprintf("const EXPECTED_INTERCEPT_PLS4: f64 = %.10f;\n", intercept4))
cat("#[rustfmt::skip]\n")
cat(sprintf("const EXPECTED_COEFS_PLS4: [f64; %d] = [\n    ", p))
cat(paste(sprintf("%.10f", coefs4), collapse = ", "))
cat("\n];\n")
cat(sprintf("const EXPECTED_R2_PLS4: f64 = %.10f;\n", r2_4))
cat("\n")

# -----------------------------------------------------------------------------
# Summary
# -----------------------------------------------------------------------------
cat("// =============================================================================\n")
cat("// Summary of expected values\n")
cat("// =============================================================================\n")
cat(sprintf("// Test 1: n=%d, p=%d, ncomp=%d, R²=%.6f\n", 30, 5, 2, r2_1))
cat(sprintf("// Test 2: n=%d, p=%d, ncomp=%d, R²=%.6f\n", 50, 8, 4, r2_2))
cat(sprintf("// Test 3: n=%d, p=%d, ncomp=%d, R²=%.6f (with scaling)\n", 40, 6, 3, r2_3))
cat(sprintf("// Test 4: n=%d, p=%d, ncomp=%d, R²=%.6f\n", 15, 3, 1, r2_4))