Welcome to the HowToFit Jupyter Notebook lectures!
At the core of data science is fitting a model to data. This process extracts meaningful patterns, relationships, and insights, enabling accurate predictions, decision-making, and understanding of underlying processes.
However, data science can be quite challenging. With a vast array of statistical methods to choose from, it can be difficult to determine the right one for your problem. Interpreting large volumes of results is complex, and managing big datasets requires significant computational power and sophisticated statistical methods.
The HowToFit lectures teach you how to perform effective data science analysis. Designed at an undergraduate level, these lectures assume no prior knowledge of model-fitting, Bayesian statistics, or scientific analysis. They introduce core concepts without formal statistical equations, aiming to provide an understanding of the phenomenological methods used in data science. By the end of the lectures, you'll be equipped to perform your own data analysis.
The lectures use the probabilistic programming language PyAutoFit, an open-source library for model-fitting, scientific analysis, and big data analysis (https://github.com/rhayes777/PyAutoFit).
For the STFC Summer School, we will cover a subset of the HowToFit lectures from chapters 1:
Chapter 1: Introduction: How to fit a model to data, perform statistical inference, and interpret the results for scientific analysis.
More chapters are currently being written (but are not yet complete), however if you find the workshop useful reach out to me for information on the full set of lectures.
__Chapter 1: Introduction__
The first chapter of the HowToFit lectures covers the basics of model-fitting, statistical inference, and scientific interpretation. The chapter includes:
tutorial_1_models.py: What probabilistic models are and how to compose them using PyAutoFit.
tutorial_2_fitting_data.py: Fitting a model with an input set of parameters to data and quantifying the goodness of fit.
tutorial_3_non_linear_search.py: Searching non-linear parameter spaces to find the best-fit model.
tutorial_4_why_modeling_is_hard.py: Why modeling becomes difficult and how to over model-fitting problems.
tutorial_5_astronomy_examples.py: Applying the techniques to a real-world astronomy problem.
The lectures can be run via a Binder web server by using the URL below.
Tests performed for this URL were unreliable on Tuesday 16th July, whereby the binder was unable to build and produced errors. If you encounter an error, try refreshing the page in your web browser.
However, if the URL does not build successfully follow the local installation instructions below.
https://mybinder.org/v2/gh/Jammy2211/autolens_workspace/release
We strongly recommend that you install PyAutoFit in a Python virtual environment, with the link attached describing what a virtual environment is and how to create one.
The latest version of PyAutoFit is installed via pip as follows (specifying the version as shown below ensures the installation has clean dependencies):
pip install autofitIf this raises no errors PyAutoFit is installed!
Next, clone the autofit workspace (the line --depth 1 clones only the most recent branch on
the autofit_workspace, reducing the download size):
cd /path/on/your/computer/you/want/to/put/the/autofit_workspace
git clone https://github.com/Jammy2211/stfc_data_summer_school_2024 --depth 1
cd autofit_workspaceThe workspace contains the HowToFit lectures, which are Jupyter notebooks that can be run using your standard Jupiter notebook environment.