Data Summer Academy

In this first theoretical workshop, we will kick off the program with an introduction to Python and Pandas, building a parallel with Excel on how to create columns, summarise information and retrieve quick statistics. Questions, such as “how can we process data with code?” and “which issues should I be addressing with data?” will be explored, through guided examples on a dataset to answer business questions.

In the second part, examples on Pandas built-in data visualisation quick plots will be provided and we will learn about selecting the correct types of plots and how to visualise them. Additionally, we will be using Seaborn to create data visualisations that enable us to show the answers from the analysis, in a clear and objective manner.

The second module of the program’s first week will progress into the topic of Data Sourcing. We will explore different sources available to store data, such as CSV and Excel files, databases and others, as well as cover other related topics such as retrieving data through Structured Query Language (”SQL”), Application Programming Interfaces (”API”) and Web Scraping, providing both the theory on what exactly are these methods and how to use them, as well as showcasing practical examples of their applicability.

The project

The research aimed at characterizing how people move within a municipality, and to identifying intervention priorities – for instance, areas where movement restrictions are less effective. In practice, it aimed at implementing actions such as awareness campaigns or management of accesses to certain areas, with a specific focus on Cascais and Lisbon.

The first step was to conduct a cluster analysis classifying each statistical section of the municipality according to their main land use. We then analyzed anonymized data regarding counts of distinct mobile devices that connected to the network within each of the statistical sections. 

About Susana

Susana Lavado is a social psychologist turned data scientist for the love of data. She holds a PhD in Social Psychology from the University of Lisbon. After her PhD, she completed a Masters in Data Science and Advanced Analytics at Nova Information Management School. Previously, she worked in the coordination of international comparative surveys in Portugal (e.g., European Social Survey), and as a data scientist at NOS.

Susana is currently exploring the role of machine learning in social organizations, working in a range of topics with a social good focus. Her main projects include analyzing fishing patterns in Portugal during the lockdown, predicting school failure, and analyzing people’s mobility during the pandemic.

Week 2 will be moving into more advanced topics such as Machine Learning and Deep Learning, preserving a beginner-friendly approach. In the first workshop of the program’s second stage, participants will learn what Machine Learning is and its main applications on the present day.

Additionally, the concepts of “supervised” and “unsupervised” learning will be explored, as well as “classification” and “regression”, and the different metrics that shall be analysed. Finally, an introduction to models, such as Linear Regressions, Support-Vector Machines (”SVM”), KNN algorithms (for classification and regression) and Decision Trees will be performed, closing the session with a talk about the existence of more complex algorithms.

The final workshop of the program will cover what Deep Learning is and its current applications, explaining the concept of “neuron” (in linear regressions and non-linear functions) and “Neural Networks” as a sequence of neurons. In the latter, we will learn about the application of loss function and propagation, with an example of a weight update being provided.

Furthermore, we will learn about the pros and cons of Deep Learning, showing the high performance of current models and reflecting on lack of explainability, memory and time consumption. To finalize the session, we will demonstrate how to code a Neural Network in Python using a practical example.

The project

The proliferation of smartphones catalyzes a global phenomenon of "camera eats first", i.e., taking and sharing food images on social media when customers are dining out in restaurants has become increasingly popular as a key component of their life experiences. Therefore, understanding how restaurants are portrayed on social media is imperative not only for formulating adequate social media business strategies and operational actions, but also for designing effective management of customer gastronomic experiences. In this project, we aim to explore sensory gastronomic experiences portrayed on social media using a large-scale dataset of food images collected from social media platform. We propose a food aesthetics assessment model using computer vision and deep learning techniques to estimate aesthetic scores of food images. We evaluate the degree of agreement between our model's predicted aesthetics scores and human raters' aesthetic preferences of food images.

About Qiwei

Qiwei Han is currently an Assistant Professor of Data Science and Business Analytics at Nova School of Business and Economics (Nova SBE), Portugal. He received Ph.D. in Engineering and Public Policy and M.S. in Information Networking from Carnegie Mellon University. His research is at the intersection of econometrics and machine learning, using complex data-driven approaches on a variety of projects with societal impacts. He also served as the Technical Mentor for Data Science for Social Good Europe program in 2017 and 2018.