Commit | Line | Data |
---|---|---|
43a6578d BA |
1 | standardCV_core <- function(data, target, task = NULL, gmodel = NULL, params = NULL, |
2 | loss = NULL, CV = list(type = "MC", V = 10, test_size = 0.2, shuffle = TRUE) | |
3 | ) { | |
4 | if (!is.null(task)) | |
5 | task = match.arg(task, c("classification", "regression")) | |
6 | if (is.character(gmodel)) | |
7 | gmodel <- match.arg(gmodel, c("knn", "ppr", "rf", "tree")) | |
8 | if (is.numeric(params) || is.character(params)) | |
9 | params <- as.list(params) | |
10 | if (is.null(task)) { | |
11 | if (is.numeric(target)) | |
12 | task = "regression" | |
13 | else | |
14 | task = "classification" | |
15 | } | |
16 | ||
17 | if (is.null(loss)) { | |
18 | loss <- function(y1, y2) { | |
19 | if (task == "classification") { | |
20 | if (is.null(dim(y1))) | |
21 | mean(y1 != y2) | |
22 | else { | |
23 | if (!is.null(dim(y2))) | |
24 | mean(rowSums(abs(y1 - y2))) | |
25 | else { | |
26 | y2 <- as.character(y2) | |
27 | names <- colnames(y1) | |
28 | positions <- list() | |
29 | for (idx in seq_along(names)) | |
30 | positions[[ names[idx] ]] <- idx | |
31 | mean(vapply( | |
32 | seq_along(y2), | |
33 | function(idx) sum(abs(y1[idx,] - positions[[ y2[idx] ]])), | |
34 | 0)) | |
35 | } | |
36 | } | |
37 | } | |
38 | else | |
39 | mean(abs(y1 - y2)) | |
40 | } | |
41 | } | |
42 | ||
43 | n <- nrow(data) | |
44 | shuffle_inds <- NULL | |
45 | if (CV$type == "vfold" && CV$shuffle) | |
46 | shuffle_inds <- sample(n, n) | |
47 | get_testIndices <- function(v, shuffle_inds) { | |
48 | if (CV$type == "vfold") { | |
49 | first_index = round((v-1) * n / CV$V) + 1 | |
50 | last_index = round(v * n / CV$V) | |
51 | test_indices = first_index:last_index | |
52 | if (!is.null(shuffle_inds)) | |
53 | test_indices <- shuffle_inds[test_indices] | |
54 | } | |
55 | else | |
56 | test_indices = sample(n, round(n * CV$test_size)) | |
57 | test_indices | |
58 | } | |
59 | list_testinds <- list() | |
60 | for (v in seq_len(CV$V)) | |
61 | list_testinds[[v]] <- get_testIndices(v, shuffle_inds) | |
62 | ||
63 | gmodel <- agghoo::Model$new(data, target, task, gmodel, params) | |
64 | best_error <- Inf | |
65 | best_model <- NULL | |
66 | for (p in seq_len(gmodel$nmodels)) { | |
67 | error <- 0 | |
68 | for (v in seq_len(CV$V)) { | |
69 | testIdx <- list_testinds[[v]] | |
70 | dataHO <- data[-testIdx,] | |
71 | testX <- data[testIdx,] | |
72 | targetHO <- target[-testIdx] | |
73 | testY <- target[testIdx] | |
74 | if (!is.matrix(dataHO) && !is.data.frame(dataHO)) | |
75 | dataHO <- as.matrix(dataHO) | |
76 | if (!is.matrix(testX) && !is.data.frame(testX)) | |
77 | testX <- as.matrix(testX) | |
78 | model_pred <- gmodel$get(dataHO, targetHO, p) | |
79 | prediction <- model_pred(testX) | |
80 | error <- error + loss(prediction, testY) | |
81 | } | |
82 | if (error <= best_error) { | |
83 | newModel <- list(model=model_pred, param=gmodel$getParam(p)) | |
84 | if (error == best_error) | |
85 | best_model[[length(best_model)+1]] <- newModel | |
86 | else { | |
87 | best_model <- list(newModel) | |
88 | best_error <- error | |
89 | } | |
90 | } | |
91 | } | |
92 | best_model[[ sample(length(best_model), 1) ]] | |
93 | } | |
94 | ||
95 | standardCV_run <- function( | |
96 | dataTrain, dataTest, targetTrain, targetTest, verbose, CV, floss, ... | |
97 | ) { | |
98 | s <- standardCV_core(dataTrain, targetTrain, ...) | |
99 | if (verbose) | |
100 | print(paste( "Parameter:", s$param )) | |
101 | ps <- s$model(test) | |
102 | err_s <- floss(ps, targetTest) | |
103 | if (verbose) | |
104 | print(paste("error CV:", err_s)) | |
105 | invisible(c(errors, err_s)) | |
106 | } | |
107 | ||
108 | agghoo_run <- function( | |
109 | dataTrain, dataTest, targetTrain, targetTest, verbose, CV, floss, ... | |
110 | ) { | |
111 | a <- agghoo(dataTrain, targetTrain, ...) | |
112 | a$fit(CV) | |
113 | if (verbose) { | |
114 | print("Parameters:") | |
115 | print(unlist(a$getParams())) | |
116 | } | |
117 | pa <- a$predict(dataTest) | |
118 | err <- floss(pa, targetTest) | |
119 | if (verbose) | |
120 | print(paste("error agghoo:", err)) | |
121 | } | |
122 | ||
123 | # ... arguments passed to agghoo or any other procedure | |
124 | compareTo <- function( | |
125 | data, target, rseed=-1, verbose=TRUE, floss=NULL, | |
126 | CV = list(type = "MC", | |
127 | V = 10, | |
128 | test_size = 0.2, | |
129 | shuffle = TRUE), | |
130 | method_s=NULL, ... | |
131 | ) { | |
132 | if (rseed >= 0) | |
133 | set.seed(rseed) | |
134 | n <- nrow(data) | |
135 | test_indices <- sample( n, round(n / ifelse(n >= 500, 10, 5)) ) | |
136 | trainData <- as.matrix(data[-test_indices,]) | |
137 | trainTarget <- target[-test_indices] | |
138 | testData <- as.matrix(data[test_indices,]) | |
139 | testTarget <- target[test_indices] | |
140 | ||
141 | # Set error function to be used on model outputs (not in core method) | |
142 | if (is.null(floss)) { | |
143 | floss <- function(y1, y2) { | |
144 | ifelse(task == "classification", mean(y1 != y2), mean(abs(y1 - y2))) | |
145 | } | |
146 | } | |
147 | ||
148 | # Run (and compare) all methods: | |
149 | runOne <- function(o) { | |
150 | o(dataTrain, dataTest, targetTrain, targetTest, verbose, CV, floss, ...) | |
151 | } | |
152 | if (is.list(method_s)) | |
153 | errors <- sapply(method_s, runOne) | |
154 | else if (is.function(method_s)) | |
155 | errors <- runOne(method_s) | |
156 | else | |
157 | errors <- c() | |
158 | invisible(errors) | |
159 | } | |
160 | ||
161 | # Run compareTo N times in parallel | |
162 | compareMulti <- function( | |
163 | data, target, N = 100, nc = NA, | |
164 | CV = list(type = "MC", | |
165 | V = 10, | |
166 | test_size = 0.2, | |
167 | shuffle = TRUE), | |
168 | method_s=NULL, ... | |
169 | ) { | |
170 | if (is.na(nc)) | |
171 | nc <- parallel::detectCores() | |
172 | compareOne <- function(n) { | |
173 | print(n) | |
174 | compareTo(data, target, n, verbose=FALSE, CV, method_s, ...) | |
175 | } | |
176 | errors <- if (nc >= 2) { | |
177 | require(parallel) | |
178 | parallel::mclapply(1:N, compareOne, mc.cores = nc) | |
179 | } else { | |
180 | lapply(1:N, compareOne) | |
181 | } | |
182 | print("Errors:") | |
183 | Reduce('+', errors) / N | |
184 | } | |
185 | ||
186 | # TODO: unfinished ! |