You have received an e-mail from a good colleague.
We have prepared a notebook example to show an exploratory analysis of a moving cluster using data from the Gaia satellite. You can follow the example in different formats:
- See how the final notebook looks like in github: exploratory Gaia analysis
- Execute and interact in a life version in Binder: Binder exploratory Gaia analysis
- Try the notebook on your local machine (see Quick Start for installation instructions).
More details can be found in this tutorial
- Jupyter notebooks as a dynamic tool for exploratory analysis
- Initialize a notebook
- Basic structure and syntax: cells
- droplets resources
- fortran magic
- Server options
- pros
- Felixibility for exploratory data, training, sharing
- Web app, accessible from anywhere (ssh, server)
- Markdown + code + resuts. Science results are more "tangible"
- Reports in different formats, dashboards
- Many extensions, and growing!
- cons
- Hidden state and out-of-order execution
- Notebooks encourage bad habits (not ideal for software development)
- In general, not as powerful as a stand-alone application or modules (not ideal for sharing good code)
- Some difficulties to obtain diff
- Jake VanderPlas youtube series on Reproducible data analysis with jupyter Youtube
- Try Jupyter in your browser link
- Quickview Notebook sharing the Gravitational Wave detection Notebook
- A Machine Learning course using Notebooks: Lecture 1: Density Est, Lecture 3: Classification and Lecture 4: Dimensionality Reduction.
- The full tutorial on an international Python conference: PyCon 2015 Scikit-learn Tutorial
- Easy to learn tool
- Interweave results, ideas, and hypotheses with the code
- Natural format to create a scientific narrative
- State of scripts is not linear, depends on user
- Excellent tools to share your research
- March 16
- Collaborative Jupyter notebooks through GitHub