Foundations: Getting started with engineering data
These materials establish the shared computational and statistical vocabulary used throughout MECE 4520, using the course’s recurring gas-turbine emissions dataset. Complete the first three notebooks in order before beginning the course’s main modules; plan for approximately 2.5–3 hours in total. The CLT interactive is a companion to the first in-class discussion and can be revisited as needed.
We highly recommend working with Google Colab. This allows you to run the notebooks in the cloud without tedious local setup. At the end of each notebook, there is a “Check-in” section that encourages you to try out what you’ve learned.
The notebooks use the course version of the gas-turbine dataset. It retains the original measurements while adding a campaign_year field and documented missing values for later data-pipeline exercises. If opened independently in a fresh environment, they download that same course version from this repository.
1. Course tools and Python foundations
Estimated time: 30–45 minutes
Learn the minimal NumPy and Pandas workflow used in the course: loading data, selecting variables, filtering observations, and making a first plot.
2. A first look at gas-turbine emissions data
Estimated time: 45–60 minutes
Explore the measurements, variables, distributions, and simple physical questions in the Gas Turbine CO and NOx Emissions dataset.
3. Probability distributions and simulation
Estimated time: 45–60 minutes
Practice drawing and visualizing simulated normal, uniform, and exponential data, and connect their shapes to familiar and engineering situations.
4. Central Limit Theorem: interactive simulation and notebook
Estimated time: 20–30 minutes
Build a distribution of sample means one batch at a time, then see how the sample size changes its shape and spread. The interactive and companion notebook accompany the first in-class discussion of the central limit theorem.
5. Confidence intervals: interactive coverage simulation
Estimated time: 10–15 minutes
Repeatedly draw samples from a known normal population and construct a 95% confidence interval for each. See why 95% refers to the long-run coverage of the procedure, rather than a guarantee for one interval.
Dataset source
The underlying measurements come from the Gas Turbine CO and NOx Emissions dataset at the UCI Machine Learning Repository. The source data contain 36,733 hourly measurements from a gas turbine collected between 2011 and 2015 and are made available under the CC BY 4.0 license. The clean UCI files are not stored in this repository.