This Training academy aims to provide students and professionals from different fields with the technical and scientific skills to analyse, model and extract knowledge from data in a variety of business contexts, using the KNIME Analytics platform. With a strong practical approach, the program is designed to present, discuss, and constructively apply classical and state-of-the-art approaches for descriptive and predictive analytics as well as for time series analysis and forecasting. Throughout the program, the participants are exposed to real world case studies, with varying degrees of complexity, for which they are challenged to develop, test, and evaluate analytical models able to extract useful knowledge and actionable insights from the data.
The main learning outcome is to enable students to gain proficiency in using KNIME as a white-box tool for modelling and implementing transparent and reproducible data workflows in an automatic fashion, towards proactive, data-driven decision-making.
Overview
The program starts by introducing the fundamentals of Analytics and its applications, as well as the KNIME Analytics platform and its main functionalities, so as to provide context and motivation for the students. The remainder of the program is structured in three main parts. Firstly, the students are introduced to descriptive analytics, in which they are exposed to and apply exploratory data analysis techniques, including pre-processing and transformation of variables in a data frame. Unsupervised learning techniques are also explored and applied, namely in the context of dimensionality reduction, clustering, and association. Secondly, the program covers the area of predictive analytics, with particular emphasis on understanding, implementing, and evaluating regression and classification models. Finally, some topics on the analysis, modelling and forecasting of uni/multivariate time series are covered, using statistical and machine learning techniques.
The teaching methodology follows a strategy with a strong practical component, in which the introduction of the various contents is supported by the discussion and resolution of case studies. Here, students are expected to develop, test, and critically analyse data pipelines based on real datasets of different size and structure.
Who can attend?
This Training academy is mainly targeted at graduate students (MSc and PhD) and professionals in quantitative scientific fields that are not familiar with programming languages but are interested in developing technical skills in data mining, particularly:
- Students and professionals from STEM and non-STEM areas, with a basic analytical background, who are interested in data mining.
- Data/Business analysts interested in consolidating their knowledge in business analytics;
- Software engineers interested in the development of machine learning models and pipelines.
As entry requirements, it is assumed that the student has basic knowledge of manipulating and statistically exploring datasets using conventional spreadsheets.
Programme Outline
The syllabus has been designed and structured to be gradually more complex, allowing the students to better understand and connect the concepts learned.
The main objective of this programme is to explore the KNIME Analytics Platform tool in various case studies, based on the theoretical content learned at the beginning of each module. It is intended that students acquire basic knowledge of data mining and modeling.
Day 1: Foundations of Analytics
Topics: Fundamental concepts and applications; Problem solving with analytics
Day 2: Introduction to KNIME Analytics platform
Topics: Basic terminology, settings and user interface; Workflow control; Data workflows in KNIME workbench
Day 3: Hands-on Descriptive Analytics with KNIME
Topics: Exploratory data analysis: descriptive statistics and data visualization; Data pre-processing and data transformation; Introduction to unsupervised learning
Day 4: Hands-on Predictive Analytics with KNIME
Topics: Introduction to supervised learning; Regression and classification models; Classical and robust evaluation of predictive models
Day 5: Time series analysis and forecasting with KNIME
Topics: Data preparation; Time series analysis and visualization; Time series forecasting as a supervised learning problem; Uni/multivariate time series forecasting; Forecasting evaluation
Teaching
João Gonçalves holds a PhD in Industrial and Systems Engineering, a MSc in Systems Engineering and a BSc in Mathematics from the University of Minho (UMinho). He is a Lecturer in Data Science and Coordinator of the BSc in Applied Data Science at the Catholic University of Portugal (UCP). He also works as an Invited Assistant Professor at UMinho in the areas of Operations Research and Supply Chain Management. He has worked as Principal Data Scientist at Bosch Car Multimedia and Technical Coordinator of the Data Science & Machine Learning Group of the Collaborative Laboratory on Digital Transformation (DTx). His research interests lie in the areas of Machine Learning, Forecasting and Business Analytics.