eric-czech

# Parallel Cross-Validation Example

library(foreach)    # install.packages('foreach')
library(caret)      # install.packages('caret', dependencies = c("Depends", "Suggests"))
library(doParallel) # install.packages('doParallel')
registerDoParallel(makeCluster(4)) # Use 4 cores for parallel CV

data <- read.csv(…) # Assuming this is your dataset 

cv <- caret::createFolds(nrow(data), k=10, list=T) # Create 10 folds

eric-czech

# As an administrative user, run the following on command line (paths need to be adjusted to local jre version):
root> export JAVA_HOME='/Library/Java/JavaVirtualMachines/jdk1.7.0_91.jdk/Contents/Home/jre'
root> export LD_LIBRARY_PATH='/Library/Java/JavaVirtualMachines/jdk1.7.0_91.jdk/Contents/Home/jre/lib/server'
root> export PATH=$PATH:$JAVA_HOME/bin
root> R CMD javareconf JAVA_CPPFLAGS=-I/System/Library/Frameworks/JavaVM.framework/Headers

# On the R command line (launched as admin), run the following
R> install.packages('rJava','http://www.rforge.net/')

eric-czech

# These instructions were used to create a CUDA 9.1 compatible EC2 instance for the sake of 
# running experiments with [imagej-ops-experiments]([ops-experiment](https://github.com/imagej/ops-experiments) 
# deconvolution algorithms (via YacuDecu)

# Step 1) Create Instance
# Start instance for image: https://aws.amazon.com/marketplace/pp/B07BFSMQ32
# - Choose p2.xlarge type
# - Make sure to have at least 30G of disk space

# Step 2) Install Java/Maven

eric-czech

#' Get density associated with points in 2D space
#' 
#' @param x Vector of numeric values
#' @param y Vector of numeric values; must have equal length to \code{x}
#' @param nbins Grid size used for 2D kde. Larger numbers mean density bins are 
#'   more continuous and at around 50 visible discretization begins to disappear 
#'   in scatterplots (though lower numbers will improve speed)
#' @param limit Maximum number of points to use for kde estimation
#' @param ... Other arguments for MASS::kde2d
#' @return A vector of densities with length equal to length of \code{x} and \code{y}.  

eric-czech

RUN mkdir $HOME/.nodeenv && \
    cd $HOME/.nodeenv && \
    pip install nodeenv && \
    nodeenv jupyterlab && \
    . jupyterlab/bin/activate && \
    pip install itkwidgets && \
    jupyter labextension install @jupyter-widgets/jupyterlab-manager itk-jupyter-widgets

eric-czech

from sklearn.model_selection import train_test_split
import numpy as np
    
    
def _get_stratified_split(splits, idx, values, proportions):
    # Return immediately if there are fewer than 2 proportions use in splits
    if len(proportions) <= 1:
        splits.append(idx)
        return splits
    

eric-czech

import numpy as np
from io import BytesIO
from PIL import Image

def png_bytes_to_numpy(png):
    """Convert png bytes to numpy array
    
    Example:  
    
    >>> fig = go.Figure(go.Scatter(x=[1], y=[1]))

eric-czech

eric-czech

eric-czech