Abstract
Air pollution kills approximately 7 million people every year and nine out of ten people are exposed to high levels of airborne pollutants. This paper describes the design of a bicycle air probe by a team of multicultural and multidisciplinary students of the European Project Semester, during the spring of 2020. This learning experience started with the analysis of the state-of-the-art, ethics, marketing and sustainability dimensions, and was followed by the design, development and simulation of a proof-of-concept solution. The result is GOairLight - a bicycle probe paired with a mobile app. The probe collects air quality, humidity and temperature data as cyclists ride, while the mobile app shares the collected data with the community, by means of a cloud database, presents relevant air quality information and suggests less polluted routes. Furthermore, it relies on a sustainable energy source - a dynamo powered by the cyclist - and automatic lighting. The latter feature improves cyclist visibility and raises the awareness towards the cyclist, contributing to increased road safety.

Abstract
Crowdsourced data streams are continuous flows of data generated at high rate by users, also known as the crowd. These data streams are popular and extremely valuable in several domains. This is the case of tourism, where crowdsourcing platforms rely on tourist and business inputs to provide tailored recommendations to future tourists in real time. The continuous, open and non-curated nature of the crowd-originated data requires robust data stream mining techniques for on-line profiling, recommendation and evaluation. The sought techniques need, not only, to continuously improve profiles and learn models, but also be transparent, overcome biases, prioritise preferences, and master huge data volumes; all in real time. This article surveys the state-of-art in this field, and identifies future research opportunities. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract
In the spring of 2020, six undergraduate students from diverse countries and engineering fields decided to design together a solution to monitor the elderly. This project was performed as part of the European Project Semester (EPS) programme at Instituto Superior de Engenharia do Porto (ISEP). The EM-BRACE solution encompasses two interconnected devices (a home station and a bracelet) and mobile/Web twin applications. The bracelet measures and transmits vital user data (pulse, temperature and impacts) to the home station, whereas the latter measures home environment parameters (temperature, humidity and pressure) and sends local and bracelet data to an Internet of Things (IoT) platform. This way, these data become accessible via the mobile/Web application. Thereby, EM-BRACE monitors the health and environment of the elderly and timely notifies caregivers about problems, contributing to the well-being of the elderly and their families. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.

Abstract
The number of Internet of Things devices generating data streams is expected to grow exponentially with the support of emergent technologies such as 5G networks. Therefore, the online processing of these data streams requires the design and development of suitable machine learning algorithms, able to learn online, as data is generated. Like their batch-learning counterparts, stream-based learning algorithms require careful hyperparameter settings. However, this problem is exacerbated in online learning settings, especially with the occurrence of concept drifts, which frequently require the reconfiguration of hyperparameters. In this article, we present SSPT, an extension of the Self Parameter Tuning (SPT) optimisation algorithm for data streams. We apply the Nelder-Mead algorithm to dynamically-sized samples, converging to optimal settings in a single pass over data while using a relatively small number of hyperparameter configurations. In addition, our proposal automatically readjusts hyperparameters when concept drift occurs. To assess the effectiveness of SSPT, the algorithm is evaluated with three different machine learning problems: recommendation, regression, and classification. Experiments with well-known data sets show that the proposed algorithm can outperform previous hyperparameter tuning efforts by human experts. Results also show that SSPT converges significantly faster and presents at least similar accuracy when compared with the previous double-pass version of the SPT algorithm.

Abstract
In spite of their growing maturity, telecommunication operators lack complete client characterisation, essential to improve quality of service. Additionally, studies show that the cost to retain a client is lower than the cost associated to acquire new ones. Hence, understanding and predicting future client actions is a trend on the rise, crucial to improve the relationship between operator and client. In this paper, we focus in pay-as-you-go clients with uneven top-ups. We aim to determine to what extent we are able to predict the individual frequency and average value of monthly top-ups. To answer this question, we resort to a Portuguese mobile network operator data set with around 200 000 clients, and nine-month of client top-up events, to build client profiles. The proposed method adopts sliding window multiple linear regression and accuracy metrics to determine the best set of features and window size for the prediction of the individual top-up monthly frequency and monthly value. Results are very promising, showing that it is possible to estimate the upcoming individual target values with high accuracy.

Abstract
The European Project Semester (EPS) offered by the Instituto Superior de Engenharia do Porto (ISEP) provides engineering, business and product design undergraduates with a project-based learning experience in a multicultural and multidisciplinary teamwork environment. This paper reports the research and development of a reconfigurable and ergonomic three-level desk, for people who live in small spaces, by a multicultural and multidisciplinary team of five students. The main objective of the project was to integrate ethics- and sustainability-driven practices in the design, simulation and test an ergonomic, transformable desk. The FREE desk proposal aims to create a comfortable and dynamic working environment for people while providing a transformable space for different daily activities. This goal was pursued by designing a reconfigurable product, a smart desk that offers the user three levels of adjustability: bench level, sitting desk level, and standing desk level. The desk includes a folding light-sensor lamp into the table top and an integrated battery, in order to create a proper working space. The selected materials have a low environmental impact. The solution comes with different options regarding the table top lifting mechanism. This paper describes the state-of-the-art research, the ethics, sustainability, and marketing analyses, the design and simulation of the FREE desk as well as the obtained results.