Modelling in Advanced Data analytics

Aims of the course

The aim of this course is to introduce the students to the concepts of advanced modelling, encourage their creativity in analysis, finding solutions to the greatest challenge of in selecting the most appropriate technique and deploying it in a meaningful way. We will present the most common methods of predictive analytics and data mining on various business fields. The objective is to turn data into valuable information, enabling managers to make better decisions.

Course syllabus

1. Introduction: The role of modelling in advanced data analytics
2. Classification
a. The notion of classification and classifiers
b. Presentation and comparison of various classifiers: tree based, neural network based, parametric models
c. Applications: churn prediction, clothing size determination,…
3. Model evaluation
a. How well do the models perform: direct and cross evaluation
b. Confusion matrix, ROC and AUC comparisons, Lift curves
4. Feature selection
a. Which attributes we need for the model
b. Information theory filtering, Tree weighting
5. Models for predicting numeric outcomes
a. Applications: sales predictions, daily/hourly electricity consumption prediction, weather based sales
6. Association analysis
a. What to offer and to whom
b. Application: Shopping basket analysis and recommender automation
7. Model based segmentation
a. Segmentation models
b. Application: segmenting customers based on shopping behavior

Course director(s)