Statistical Modelling with Python

The Python programming language is one of the most popular programming languages in the world, and is leading the revolution in the data science ecosystem. It emphasises clear, readable code, supports a range of programming approaches, and has a huge array of packages to support almost any need.

While Python is a general-purpose programming language used for a wide range of tasks, the last 20 years have seen significant investment in its data handling, visualisation, and statistical capabilities. However, if you have never programmed before, using Python can be intimidating, and its barrier to entry is arguably higher than specialised statistical languages like R.

This course represents a more advanced take on the use of Python for building and interpreting statistical models. A lot of the course is focused on linear regression and its variations, but later chapters expand to factor analysis and clustering models. The course emphasises principled model building and interpretation over the use of inferential methods such as null hypothesis significance testing and more traditional frequentist inference methods. There’s also little content on data handling and preparation, but see here for more content in that realm.

I strongly recommend the use of the Anaconda distribution as well the JupyterLab environment. Anaconda will come with all of those readily installed, as well as almost all of the packages one uses to be productive with data in Python.

Please reach out with any questions or if I have made mistakes anywhere - there are always some!

Python Basics

• 1 - Introduction
• Using Python for Data Analysis - Exercises & Answers
• 1. Calculations in Python, and print
• 2. Variables and data types
• 3. Data Containers - the list
• 4. Data Containers - the dictionary
• 5. Using Pandas for some data analysis!
• 6. Plotting

The GLM

• 2 - The General Linear Model
• Getting to grips with linear models in Python - Exercises & Answers
• 1. Basic statistics with pingouin
• 2. The general linear model with statsmodels
• 3. Comparing and contrasting analyses through the GLM

Advancing the GLM

• 3 - Advancing the general linear model
• Checking assumptions and including interactions - Exercises & Answers
• 1. Fitting some models and checking our assumptions
• 2. Expanding the model to improve inferences

The art of prediction

• 4 - The art of prediction
• Asking scientific questions of models - Exercises & Answers
• Traditional approaches from a model-based perspective

Binary outcome models

• 5 - Binary outcome models
• Using and understanding probit models - Exercises & Answers
• 1. Iceberg, right ahead!
• 2. The Challenger Disaster
• 3. Marginally significant meta-scientific statistics

Applications and other kinds of hypotheses

• 6 - Precision in Predictions
• Intervention data
• Testing hypotheses
• Testing an interval hypothesis
• One more example
• Example exercises using skills so far
• Starting off
• Case Study One
• Case Study Two

Linear Mixed Effects Models

• 7 - Linear mixed effects models
• Dealing with repeated measurements - Exercises & Answers
• 1. Educational hierarchies
• 2. Cosmetics and attractiveness
• 3. Further challenges in linear mixed effects models
• 4. Studying individual differences with mixed models

Exploratory Factor Analysis

• 8 - Exploratory Factor Analysis
• Finding latent variables in data - Exercises & Answers
• 1. Grit - what is it?
• 2. The Nerdiness Scale - how many factors determine nerd status?

Confirmatory Factor Analysis

• 9 - Confirmatory Factor Analysis
• Fitting latent variables onto data - Exercises & Answers
• 1. Grit revisited - are we sure its there?
• Important
• Building a latent factor model of grit
• 2. Grit and conscientiousness
• 3. Latent regressions
• 4. Testing the presence of the Big Five - in Big Data
• 5. Exploring and confirming

Clustering

• 10 - Clustering
• An applied example - clustering cultures based on personality traits
• Fitting latent groups in data - Exercises & Answers
• 1. Applied clustering - types of fertility issues
• 2. Applied clustering - grouping penguins
• 3. Applied clustering - combining EFA and clustering