HTTPS SSH

julia_training

Code for the Kaggle Competition Getting Started with Julia for the team Data Science Saigon

Files

Tutorials (iJulia notebooks display better on github)

Julia Code

  • convert.jl ## Script to store .bmp files in data/trainResized in a single hdf5 file data/train.hdf5
  • con.jl ## alternate to convert.jl. Demonstrates breaking training data into train and test sets.
  • knn_julia_tutorial.jl Code from Kagle KNN tutorial tweaked to work with Julia 3.8
  • rf_julia_benchmark.jl Code from Kagle Julia tutorial tweaked to work with Julia 3.8
  • mocha_intro.jl core file for creating a submission for this competition, base on the Mocha Turorial
  • plot_statistics.jl Reads snapshots-cpu/statistics.jld (created by Mocha) and displays data using PyCall.
  • see_snapshots.jl Sample code for reading data in hdf5 files.

Data

  • /data/test.txt pointer to tell mocha to use the *.hdf5 files generated by the convert.jl script.
  • /data/train.txt pointer to data/train.hdf5
  • /data/sampleSubmission.csv contains all the test image numbers (all A submission)
  • /data/juliaSubmission.csv latest submission using Mocha
  • /data/juliaKNNSubmission.csv submission from KNN script