SigmaAnalytics & Tech

Statistical modelling

Anyone moving from 'do these differ' to 'what predicts this, and by how much' — regression as it is actually reported.

All levels3 modules9h of videoCertificate

3 modules from the Python for Data Analysis curriculum.

What you'll learn

By the end of this course

  • Write the linear model y = β₀ + β₁x + ε and interpret the intercept and slope in real units
  • Explain ordinary least squares (OLS) — fitting by minimising the sum of squared residuals
  • Fit a model three ways — scipy.stats.linregress, numpy.polyfit, and statsmodels OLS — and read the statsmodels summary table
  • Interpret R², coefficient standard errors, t-tests, p-values, and confidence intervals
  • Produce predictions with the correct uncertainty: confidence intervals (for the mean) vs prediction intervals (for a new observation)
  • Check the four regression assumptions (L·I·N·E) with residual diagnostic plots
  • Recognise the traps: extrapolation, influential points, and reading causation into a fit
  • Fit and interpret a multiple regression — coefficients as effects holding other predictors constant

About this course

3 modules from the Python for Data Analysis curriculum.

This course covers Linear Regression, Multiple Regression and Logistic Regression (Statsmodels).

About 9 hours of material across 3 modules, each with the dataset it teaches from and exercises with worked solutions.

Every module is a notebook you download and run yourself in JupyterLab — the point is that you execute the code, not that you watch someone else execute it. Recordings are added module by module; a module without one yet is taught in full by its notebook.

Who it's for

  • Researchers and postgraduate students who need the analysis done properly, not just done
  • Analysts moving from spreadsheets to something repeatable and auditable
  • Anyone who has followed a Python tutorial and still cannot open their own data and get an answer
  • Practitioners who can already run the basics and want Statistical modelling to a standard they can defend

What you'll need

  • A computer you can install Python and JupyterLab on — notebook 02 walks you through it, on Windows and on macOS
  • No prior programming experience for the foundations; the later tracks assume the earlier ones
  • Comfort with the earlier tracks of Python for Data Analysis, or equivalent experience
  • Your own dataset is welcome but not required — every module ships with the data it teaches from

Start Statistical modelling today

Lifetime access, on any device, with your progress saved as you go.

You’ll sign in first — your access is tied to your account.