Hello,
I use the below code to run a deep learning model:
library(h2o)
library(caret)
h2o.init()
test.h2o <- as.h2o(test)
split <- h2o.splitFrame(test.h2o, ratios = 0.9, seed = 123)
training_data <- h2o.assign(split[[1]], "train")
testing_data <- h2o.assign(split[[2]], "test")
levels(test.h2o$group) <- c("L1", "L2")
training_data$group <- as.factor(training_data$group)
testing_data$group <- as.factor(testing_data$group)
hidden_list <- as.list(rep(seq(1, 10, by = 1), each = 10), seed = 123)
hyperparameters <- list(hidden = hidden_list, epochs = seq(50, 200, by = 50), seed = 123)
grid <- h2o.grid(algorithm = "deeplearning", x = c("F1", "F2", "F3", "duration"), y = "group", training_frame = training_data, hyper_params = hyperparameters, nfolds = 5, seed = 123, mini_batch_size = 64)
Although I use seed, I get different results when I run the model many times. Any suggestions on how to get the same results every time?
Best,
George