Training Workshop
Monday and Tuesday will host tutorials in data-science methods and machine learning. We will be covering the topics listed below at an intermediate level, mostly using Python.
The idea here is to provide robust and fully-functioning examples that work on scientific data (i.e., not just finding cats in images!). The code will be well-documented and open-source. We will take you through it step-by-step and you can dig deeper at your leisure.
Bring Your Own Data
Do you have a problem that could benefit from the methods covered here? Then please bring your data to the workshop! Where practical, the organisers will help you fit your data into the necessary frameworks.
Topics
The following topics will be covered:
Introduction to machine learningThis tutorial introduces machine learning and teaches how to apply regression, clustering analysis and deep-learning to simple problems. The lesson is inspired by Software Carpentry and conducted through live-coding using python. Prerequisites: a basic understanding of python, including for-loops, if-statements, using functions and basic arithmetic. |
Dynamic Dashboards and Visualisation Using Python (parallel session)This tutorial covers how to build a ‘full stack’ web dashboard to visualise data in real-time. Technologies include: plotly dash, Flask, RESTful APIs, SSL for encryption and deploying on Linux or AWS. The lesson aims to present a recipe and modular code for building a complete application that can be re-purposed for your own work. Prerequisites: an intermediate understanding of python for data science, including basic pandas and matplotlib. |
Interactive Shiny applications in R (parallel session)Shiny makes it easy to build interactive, web-deployed applications using R. Dr Louise Ord will be running a hands-on R Shiny workshop to take you through all you need to begin developing your own Shiny dashboards and visualisations. In this workshop, you will be given example code that can be used as a template for your own application development. You will be taught the key aspects of reactive programming, how to create interactivity with built-in control widgets and external R and JavaScript tools, and methods for fully customising the design and appearance of your Shiny dashboards and visualisations. Prerequisites: Basic programming skills. A basic understanding of R would be useful but is not necessary.. |
Finding Structure in DataThis lesson presents practical examples of finding and measuring structure in multi-dimensional data. Methods include cluster-finding using K-means and Gaussian Mixture Models. Prerequisites: an intermediate understanding of python for data science, including pandas and matplotlib. |
Practical Deep-Learning for ScienceConvolutional Neural Networks (CNNs) have revolutionised the field of computer vision and are increasingly used for scientific analysis. This lesson shows how to use CNNs for scientific data analysis, leveraging examples from astronomy. Prerequisites: a basic understanding of python, including for-loops, if-statements, using functions and basic arithmetic. |