The course is designed for students interested in data analysis, machine learning, and their application to engineering problems. An engineering background and basic knowledge of statistics, numerical methods, and programming are beneficial, but not strictly required

Data-driven decision-making is becoming a crucial skill to deal with engineering systems that generate vast amounts of data from the automation system. This module aims to provide students with an understanding of the general concepts, advantages and limitations of Data-Driven methods for engineering applications.

This course covers methodologies necessary for inferring useful information and identifying underlying patterns from raw, incomplete, noisy and corrupted data that is present in real-life engineering applications. This is achieved by introducing concepts and methods used to model a wide range of systems based on available data.

The module will also provide the students with the opportunity to explore advanced solutions of data analytics such as nonlinear models, model selection and error estimation.

 

On completion of the module the student is expected to be able to achieve the following Learning Objective (LO):

·      LO1: Describe a number of models for supervised, unsupervised inference from data. Critically evaluate statistical analysis. Critically assess the fit of statistical models.

·      LO2: Assess the strength and weakness of each of these models, interpret the mathematical equations from linear algebra, statistics, and probability theory used in the learning models.

·       LO3: Implement efficient learning algorithms in the MATLAB language, applied to engineering problems.

·     LO4: Design test procedures in order to evaluate the model hyperparameters (model selection) and its error (error estimation). Develop an appropriate experimental research design for an engineering case study considering practical limitations.

Time

Monday 16/11/2026

Tuesday 17/11/2026

Wednesday 18/11/2026

Thursday 19/11/2026

Friday 20/11/2026

09:00–10:00

Regression fundamentals Problem formulation for supervised learning; definition of input and output variables; role of assumptions in regression models; engineering interpretation of data-driven models.

Model validation Training, validation, and test datasets; why model performance must be evaluated on unseen data; introduction to overfitting and underfitting.

From perceptron to neural networks Motivation for nonlinear models; perceptron as the basic computational unit; network architecture; activation functions; learning principle.

Unsupervised learning principles Motivation for learning from unlabelled data; exploratory analysis; pattern discovery; clustering and dimensionality reduction in engineering datasets.

10:00–11:00

Linear regression models Univariate and multivariate linear regression; least-squares estimation; interpretation of regression coefficients; assumptions and limitations of linear models.

Model selection Hyperparameters; validation procedures; k-fold cross-validation; comparison of alternative models; selection of the most suitable model structure.

Neural network training Loss functions; gradient-based learning; training and validation workflow; generalisation; regularisation and prevention of overfitting.

Clustering methods k-means clustering; similarity and distance measures; choice of number of clusters; cluster interpretation; engineering examples.

11:00–12:00

Model performance assessment Residual analysis; RMSE, MAE, MAPE, and R²; prediction error; graphical assessment of model quality; interpretation of results in engineering applications.

Error estimation Generalisation error; uncertainty in performance estimates; bias–variance trade-off; practical limitations of model evaluation with limited data.

Neural networks for regression Data preparation; normalisation; training workflow; prediction; performance evaluation; comparison with linear regression models.

Dimensionality reduction Principal Component Analysis; feature representation; visualisation of high-dimensional data; interpretation of principal components and explained variance.

12:00–13:00

Break

Break

Break

Break

Time

Monday 16/11/2026

Tuesday 17/11/2026

Wednesday 18/11/2026

Thursday 19/11/2026

Friday 20/11/2026

13:00–14:00

Introduction to data-driven engineering Role of data-driven methods in monitoring, modelling, control, diagnosis, optimisation, and decision support for engineering systems.

Hands-on 1: Linear regression in MATLAB Import an engineering dataset; define input and output variables; build univariate and multivariate regression models.

Hands-on 2: Validation workflow in MATLAB Split data into training, validation, and test sets; implement validation procedures; compare model performance on unseen data.

Hands-on 3: Neural networks in MATLAB Prepare data; define a simple neural network; train the model; monitor training and validation performance.

Hands-on 4: Unsupervised learning in MATLAB Apply clustering and PCA to an engineering dataset; visualise results; identify possible hidden structures.

14:00–15:00

Exploratory data analysis Data types; descriptive statistics; missing data; outliers; noisy measurements; corrupted data; visual inspection of engineering datasets.

Regression case study Fit regression models; compute predictions; analyse residuals; calculate RMSE, MAE, MAPE, and R²; assess model adequacy.

Model comparison exercise Compare alternative models using validation metrics; tune model hyperparameters; identify overfitting and underfitting.

Neural network case study Compare linear and nonlinear models; analyse prediction errors; discuss when neural networks are justified and when simpler models are preferable.

Interpretation exercise Interpret clusters and principal components; connect unsupervised learning results to engineering diagnosis, classification, and decision-making.

15:00–16:00

MATLAB data inspection exercise Import data; clean variables; visualise distributions and relationships; compute descriptive statistics; formulate a preliminary engineering modelling problem.

Regression interpretation and limitations Discuss modelling assumptions, data quality, residual patterns, extrapolation risks, and common mistakes in regression-based engineering analysis.

Reliable performance reporting Summarise validation results; report error metrics; justify model selection; discuss uncertainty and practical limitations of the selected model.

Nonlinear model interpretation Discuss overfitting, regularisation, model transparency, robustness, and the trade-off between accuracy and interpretability.

Coursework briefing Presentation of the coursework task; expected report structure; code requirements; assessment criteria; submission procedure; discussion of expectations and practical recommendations.

 

To ensure that students not only understand the theoretical concepts but can also apply them in practice, the learning objectives will be assessed through a coursework project.

 

The coursework is designed to evaluate students’ understanding of data-driven and AI-based methods, as well as their ability to develop code for solving practical engineering problems. It provides an opportunity for students to integrate theoretical knowledge, computational skills, and engineering judgement in a realistic application.

 

The coursework will be introduced and discussed on the final day of the course, Friday, 20 November 2026. During this session, students will receive a clear explanation of the coursework requirements, the engineering problem to be addressed, the expected modelling approach, and the assessment criteria.

 

After the coursework has been introduced, students will have three weeks to complete and submit their work. The expected submission deadline is therefore Friday, 11 December 2026, unless otherwise specified by the course organisation.

 

The submission will consist of two main components:

 

1.      A technical report describing the engineering problem, the modelling approach, the selected data-driven methods, the validation procedure, and the analysis of the results.

2.      Well-documented MATLAB/Pytho/R code implementing the selected methods and demonstrating the practical application of the techniques covered during the course.

 

Both the report and the code will be assessed. The final grade will reflect the students’ ability to combine theoretical understanding, computational implementation, critical analysis, and engineering interpretation.

City Rally on Sunday

Campus tour

Closing ceremony

Various activities organized during Athens Week by the student societies.