Regression & Experimental Design
Fit and check multiple regression and GLMs, design randomised, blocked and factorial experiments, and turn analyses into clear, reproducible reports for the researchers you work with.
Leads to: Data Scientist, Biochemist, Chemist, Materials Scientist, Statistician, Operations Research Analyst
Before you start
This course builds on Probability & Statistical Inference and Intro to Data Analysis with Python. Take them first if you are new to the topic.
About this course
A mid-level course for statisticians, data scientists, operations research analysts and lab scientists (chemists, biochemists, materials scientists) who plan studies and analyse their results. Module 1 covers regression modelling: multiple regression and its interpretation, residual and influence diagnostics, and generalised linear models for yes/no outcomes and counts. Module 2 covers the design of experiments: randomisation, replication and blocking, one-way ANOVA with follow-up comparisons, and 2^k factorial designs with interactions. Module 3 introduces mixed-effects models, response surface methodology and causal inference with DAGs. Module 4 is about the work around the analysis: running a consulting intake, writing methods and results sections, and keeping analyses reproducible in R or Python. Six lessons have Python exercises (NumPy runs in the browser); the rest have practical tasks or quizzes. Prerequisites: Probability & Statistical Inference and Intro to Data Analysis with Python.
Course content
Module 1
Regression Modelling
3 lessons · 2h 9mFit and interpret multiple regression, check residuals, leverage and influence, and model yes/no outcomes and counts with logistic and Poisson GLMs.
- 37mMultiple regressionReading
Fit a regression with several predictors and interpret each coefficient, its interval and R-squared.
- 35mDiagnosticsReading
Check a regression's assumptions with residual plots and find high-leverage and influential points.
- 57mGeneralised linear modelsReading
Model yes/no outcomes and counts with logistic and Poisson GLMs and interpret odds and rate ratios.
Module 2
Designing Experiments
3 lessons · 2h 10mPlan experiments with randomisation, replication and blocking, analyse them with one-way ANOVA and follow-up comparisons, and estimate main effects and interactions in 2^k factorial designs.
- 42mRandomisation, replication, blockingReading
Apply the three principles of experimental design and lay out a randomised complete block experiment.
- 46mANOVAReading
Compute and interpret a one-way ANOVA table and follow it with the right multiple comparisons.
- 42mFactorial designsReading
Estimate main effects and interactions from a 2^k factorial experiment and interpret them.
Module 3
Further Methods
3 lessons · 1h 32mModel clustered and repeated data with random effects, find optimal settings with response surface designs, and reason about cause and effect with DAGs.
- 39mMixed modelsReading
Explain fixed and random effects and when clustered or repeated data need a mixed model.
- 33mResponse surfacesReading
Fit a second-order response surface from a central composite design and locate its optimum.
- 20mCausal inference introReading
Draw a causal DAG, spot confounders, mediators and colliders, and choose what to adjust for.
Module 4
Statistical Consulting
3 lessons · 2h 9mRun a consulting intake with researchers, write clear methods and results sections, and keep analyses reproducible across R and Python.
- 34mWorking with researchersReading
Run a statistical consulting intake that pins down the question, the data and the deliverable.
- 35mStatistical reportsReading
Write a reproducible methods section and a results section that leads with effect sizes and uncertainty.
- 1hStatistical software and reproducibilityReading
Compare R and Python for statistics and rerun an analysis reproducibly in a second tool.