SmartTensors workshop materials

Contents

The Workshop will showcase how to code and perform machine-learning analyses in Julia. Especially, we will demonstrate what tools are good for geoscientific data analysis. We will cover a series of general, frequently-solved ML tasks such as:

• Unsupervised and supervised machine learning

The unsupervised machine learning methods are:

• k-means clustering
• principal component analysis
• non-negative matrix factorization
• non-negative matrix factorization with customized k-means clustering

The supervised machine learning methods include:

• neural networks
• convolutional neural networks

We will also demonstrate guided tutorials using real data

• Spatiotemporal data analytics
• GeoThermal exploration
• Filling data gaps

Getting Started

Download and install the latest version of Julia.

Julia REPL

Julia REPL looks like this:

Julia and GIT

Julia uses GIT for package management. GIT needs to be installed and configured as well. `

Jupyter Notebooks

Jupyter notebooks are in-browser interactive programming environments that we will use for this workshop. The notebooks are run through IJulia.

To install IJulia separately, open a Julia REPL and run:

	ENV["PYTHON"] = ""
	import Pkg
	Pkg.add("IJulia")

To open a Jupyter Notebook session in your browser, run the following in a REPL:

	import IJulia
	IJulia.notebook()

The first time you run this, it will install jupyter using conda.

The Workshop has the following notebooks (in the notebook directory) that can be executed in Jupyter or in the Julia REPL.

• Julia_Introduction
• Unsupervised_ML
• Supervised_ML
• Geothermalcloud

Required packages/modules to run the above-mentioned notebooks

	ENV["PYTHON"] = ""
	import Pkg
	Pkg.add("NMF")
	Pkg.add("Clustering")
	Pkg.add("Statistics")
	Pkg.add("MultivariateStats")
	Pkg.add("Distances")
	Pkg.add("Random")
	Pkg.add("Mads")
	Pkg.add("NMFk")
	Pkg.add("DelimitedFiles")
	Pkg.add("Cairo")
	Pkg.add("Fontconfig")
	Pkg.add("Gadfly")
	Pkg.add("Compose")
	Pkg.add("GMT") # this is not mandatory but an alternative tool of ArcGIS

Julia support

The official Julia documentation is available at https://docs.julialang.org.

The official Julia discourse https://discourse.julialang.org is an excellent resource for all kinds of questions and insights in addition to Stack Overflow.

In the Julia community, it is not recommended to push/pull requests, submit coding issues, or ask questions before you have checked for existing answers or insights at the Julia discourse website.