predict-cumulative.Rmd

```{r init}
source("lib/function_library.R", echo = F)

# Load other R source files in the lib directory.
load_all_code("lib")

# Load a bunch of required packages, installing missing ones as needed.
load_all_packages(auto_install = T)

# Set some standard settings.
conf = list(data_dir = "data", dir = ".")
```

```{r predict-setup}
# Setup multinode processing if available.
# Just use multicore for now.
conf$cluster = ck37r::parallelize(allow_multinode = F)#, type = "doSNOW")

#conf$method = "method.AUC"
conf$method = "method.NNLS"

conf$outer_cv_folds = 20
#conf$outer_cv_folds = 5
conf$cv_folds = 10
#conf$cv_folds = 8

# If the cluster object is null it means we're setup for multicore (or nuthin).
if (is.null(conf$cluster) || is.na(conf$cluster)) {
  # Use non-CV if we're running this in RStudio on a laptop.
  if (F && .Platform$GUI == "RStudio") {
    cat("Using normal SuperLearner.\n")
    fn = "sl_fn"
  } else {
    cat("Using CV.SuperLearner.\n")
    fn = "cv_sl_fn"
  }
  if (T) { 
    # Multicore version.
    conf$sl_fn = ckTools::gen_superlearner(parallel = "multicore",
                  outer_cv_folds = conf$outer_cv_folds)[[fn]]
  } else {
    # Sequential version.
    conf$sl_fn = SuperLearner
  }
  rm(fn)
} else {
  # Multinode version.
  conf$sl_fn = ck37r::gen_superlearner(parallel = "snow",
                  cluster = conf$cluster,
                  outer_cv_folds = conf$outer_cv_folds)$cv_sl_fn
}
```

```{r predict}
load(paste0(conf$data_dir, "/create-dataset.RData"))

if (foreach::getDoParName() == "doSNOW") {
  # Set multinode-compatible seed.
  clusterSetRNGStream(conf$cluster, iseed = 1)
} else {
  # Set multicore-compatible seed.
  set.seed(1, "L'Ecuyer-CMRG")
}

X = dataset$X
Y = dataset$outcomes$progression

libs = cumulative_library_seq(dim(X), type = "regression", glm = F)
length(libs)

#libs = libs[4:5]
#libs = libs[c(1, 2, 3, 4, 5, 7)]

results = fit_model_libs(Y, X, libs,
                         family=gaussian(), sl_fn = conf$sl_fn,
                         cv_folds = conf$cv_folds, cluster = conf$cluster)

save(results, libs, conf, dataset,
     file = paste0(conf$data_dir, "/predict_results.RData"))
```

Process results.

```{r predict-results}
load(paste0(conf$data_dir, "/predict_results.RData"))

# Create plots and generate tables for all results.
# TODO: update this function to support RMSE rather than AUC as the metric.
# Also display R^2 in this analysis.
if (F) {
plot_all_results(results, dataset$outcomes$progresson, "PD", libs,
                 type = "regression",
                 plot_dir = paste0(conf$dir, "/visuals/"),
                 tex_dir = paste0(conf$dir, "/tex/"))
}

```

```{r cleanup}
ck37r::stop_cluster(conf$cluster)
```