Hackathon projects

List of hackathon projects

Hackathon projects solve real problems. Working as a team, use Data Science to solve challenges posed by partner companies.

Hackathon projects are an essential component of the Green Data Science master programme. Data Science problems proposed by companies or entities are worked on by teams of at least five students, providing real work experience. The word Hackathon comes from hacker+marathon, which means challenge to reach the goal!

Benefits of Hackathon Projects

  • work with real problems
  • train and apply acquired skills to complex problems
  • work in collaborative teams

I want to propose a data science problem!

If, at your company or entity, you have a data science problem, and you would like it to be solved by one of our hackathon teams, you can become a partner company or entity of the Green Data Science. See how at PARTNERSHIPS.

List of Hackathon projects

The following projects were developed by the Hackathon teams

Thumbnail

 

AGROGES Quercus Project - Forest Inventory Optimization

Partner: AGROGES
Domain: Forestry
Year: 2024
Forest management plays a vital role in ensuring sustainable ecosystems and resources, yet traditional forest inventory processes are often time-consuming and resource-intensive. Forest inventories, which involve gathering and analyzing data on forest areas, are central to sustainable forest management efforts. While modern inventories have begun integrating advanced technological tools such as Geographic Information Systems (GIS), there remains a critical need to optimize these processes, particularly in regions where resources for extensive fieldwork are limited. The AGROGES company faces significant challenges in efficiently managing forest inventories for cork oak (Quercus suber) and holm oak (Quercus ilex) species. Traditional methods rely heavily on manual fieldwork and statistical analysis, making the process costly and time-consuming. The need for accurate tree counting, species differentiation, crown area calculation, and health assessment requires a more efficient and scalable approach. In this Hackathon, the challenge was to develop a Convolutional Neural Network (CNN) deep learning model capable of analyzing geospatial data to achieve multiple forest management objectives: assessing tree numbers in plots, accurately mapping the presence of both oak species, differentiating between cork oak and holm oak trees, calculating individual tree crown areas for resource assessment, and evaluating tree health and condition to inform sustainable management practices.

 

Load details

Photo by Natacha de Hepcée on Unsplash

Thumbnail

 

Wildfire Priority Prediction - Data Pipeline Enhancement

Partner: ANEPC
Domain: Wildfire management
Year: 2024
Wildfires pose a significant national challenge for Portugal due to their unpredictable nature and the multitude of variables involved. Research indicates that while most fires are controlled early, a small percentage that escape initial attack cause the majority of damage and burnt area. Fires that escape the critical initial 90-minute window require firefighters to deploy additional resources and strategies for control, often leading to larger and more devastating events. The ability to predict these scenarios is crucial for improving decision-making and resource management in real-time. Building upon the foundation established by the 2023 hackathon team, who successfully created an application that predicts which fires to prioritize using an XGBoost machine learning model, this project focused on addressing key limitations in data quantity, data pipeline quality, and prediction tool application. The challenge was particularly complex as it required handling three distinct data types: meteorological data (dynamic), spatial data (static), and fire occurrence data, while ensuring the system could be continuously updated to maintain accuracy. In this Hackathon, the team was tasked with enhancing the existing wildfire prediction system by adding new data sources, improving the data collection process for continuous model updates, and developing a secondary model that operates without first-time fire support arrival information. The project involved creating automated scripts for weekly data updates, cleaning, and preparation, with the goal of optimizing the entire pipeline from data collection to application deployment. Given that Portugal's wildfire season typically begins in July and lasts approximately 15 weeks, the enhanced system was designed to update annually before fire season, with suggested data refreshes in May.

 

Load details

Photo by Egor Vikhrev on Unsplash

Thumbnail

 

irrig.ISA Dashboard - Sustainable Water Management for Irrigation

Partner: ISA
Domain: Water management for irrigation
Year: 2024
Water management for irrigation represents one of the most critical challenges in sustainable agriculture, particularly in Portugal where water resources require careful conservation and optimization. Decades of valuable research data on water needs for key crops have been collected, but this information often remains scattered across fragmented datasets, making it difficult for farmers, irrigation technicians, and public administrators to access and apply this knowledge effectively in their daily operations. The complexity of irrigation data, combined with the diverse needs of different stakeholders, creates a significant barrier between scientific research and practical field application. Farmers need clear, actionable insights to make informed decisions about when and how much to irrigate, while maintaining sustainable practices that conserve water resources. The challenge extends beyond data access to encompass the need for intuitive tools that can translate complex scientific information into practical guidance for irrigation management. In this Hackathon, the challenge was to develop the irrig.ISA platform, a comprehensive solution that addresses these challenges through three core approaches: organizing decades of fragmented research datasets into a coherent and unified structure compatible with modern data frameworks; creating interactive dashboards and visualizations tailored to meet diverse user needs; and ensuring the platform aligns with practical requirements of end users across different sectors. The project aimed to democratize access to critical irrigation data, streamline complex datasets into visually intuitive formats, and promote sustainable resource management by enabling precise irrigation practices that reduce water waste and enhance long-term agricultural resilience.

 

Load details

Photo by Jonathan Hislop on Unsplash

Thumbnail

 

Eucalyptus Weevil Management - Defoliation Prediction and Biological Control Optimization

Partner: RAIZ - Navigator
Domain: Forestry
Year: 2024
Eucalyptus plantations represent a cornerstone of Portugal's forestry sector, with Eucalyptus globulus covering approximately 845,000 hectares and constituting 26.2% of the Portuguese forest. These plantations serve as the primary raw material source for the pulp and paper industry, one of the country's most important economic sectors. However, defoliation caused by the Eucalyptus weevil (Gonipterus platensis) poses a persistent and severe threat to plantation productivity and ecosystem health. The Eucalyptus weevil, accidentally introduced to Portugal in 1995, rapidly became a major pest with larvae and adults feeding on newly flushed leaves, shoots, and buds, causing trees to lose apical dominance and significantly decreasing wood productivity. Despite biological control efforts using the parasitoid Anaphes nitens, the effectiveness varies dramatically with environmental conditions, particularly temperature and elevation. Research shows that parasitism rates drop from 70-95% at low elevations to 0-25% above 600-700m altitude, while winter temperatures below 10°C result in ineffective biological control. The economic impact is substantial, with weevil-induced damage causing wood losses of 648 million euros over the past 20 years in Portugal. In this Hackathon, the challenge was divided into two interconnected problems addressing critical gaps in eucalyptus pest management. The first problem focused on developing predictive models for defoliation levels based on environmental and management variables, enabling decision-makers to anticipate risks, prioritize regions for monitoring and intervention, and optimize resource allocation for pest control while minimizing environmental impacts. The second problem involved analyzing temperature data from 2009-2011 to understand how meteorological factors affect population dynamics between the weevil and its parasitoid, providing insights crucial for improving biological control strategies and determining optimal conditions for natural enemy effectiveness.

 

Load details

Photo by Soliman Cifuentes on Unsplash

Thumbnail

 

Raspberry harvest forecasting challenge

Partner: The Summer Berry Company Portugal
Domain: Agriculture
Year: 2023
Predicting the quantity of fruits that can be harvested weekly is essential to support the daily management of red fruit production in greenhouses. This value is important for planning the allocation of human resources necessary to carry out the harvest, predicting the quantity of fruit delivered to customers. In the case of red fruits, which cannot be stored and have a short shelf life, the accuracy of this prediction is essential. The Summer Berry Company (TSBCo) group produces small fruits throughout the year. Production is carried out in greenhouse tunnels, which allow stable conditions for plant growth. However, not all environmental factors can be controlled, especially meteorological ones. This may have implications for the speed of fruit development and quantity produced at each time. In this Hackathon, the challenge was launched to create a short and medium-term prediction model for the quantity of raspberries harvested. The project bases its analysis on the history of data collected by the company on agricultural holdings, which includes, among others, production data, characterization of crop phenological parameters, meteorological and environmental data. Team Members Afonso Marques, Aziza Ben Tanfous, Beatriz Cardoso, Diogo Pinto, Luís Soares, Miguel Paulo Faculty advisor Rui Figueira Partner advisor João Alves, Ana Morais github repository https://github.com/isa-ulisboa/greends-hack2023-smb

 

Load details

Photo by Zach Inglis on Unsplash

Thumbnail

 

Prioritization of occurrences and allocation of means to combat forest fires

Partner: ANEPC – FEPC
Domain: Wildfire management
Year: 2023
Portugal is a country that has been severely affected by fires in recent decades, which generate high social, economic and environmental impacts. There are several factors that contribute to this reality: (1) territory with high rural abandonment, which combined with the low profitability of rural spaces, creates a flammable landscape, with high continuity and quantity of fuel; (2) favorable weather conditions, especially during summer, aggravated by climate change; (3) a high number of ignitions promoted by a culture of negligent use of fire. The fires that generate the largest burned area and negative impacts are usually concentrated in a period of 10-15 days. This fact introduces complexity into fire management, since large fires are often accompanied by simultaneity (i.e. several occurrences occurring at the same time) and high geographic dispersion across the national territory. In this context, fire suppression management is a complex task that requires knowledge, tools and concerted efforts from all involved. This must take into account the environmental conditions that influence the development of fires, but also the spatial, temporal and capacity heterogeneity of resources available for fighting. In order to improve the forest fire fighting system, it is important to increase the effectiveness of its response. This improvement may involve the promotion of anticipatory measures, which promote more effective management of resources, maintaining the balance of the combat system (i.e. avoiding its collapse). The possibility of predicting the type of event, based on the analysis of the history of occurrences, is an important element in supporting resource management decisions. In this Hackathon, the challenge is to predict the probability of a new forest fire event escaping the initial attack that allows it to be extinguished in 90 minutes, given the terrain and meteorological conditions associated with the occurrence. The aim is also to identify which factors determine a forest fire to escape this initial attack. For this project, historical data on occurrences in Portugal since 2018 is used, including operational data, meteorological data and geographic data. Team Members António Lacerda, Benjamin Hilliger, Christoph Fischer, Inês Silveira, Joana Esteves, Johanna Rauberger, Vasco Florentino Faculty advisor Akli Benali, Rui Figueira Partner advisor Fábio Silva, Alexandre Penha, Carlos Mota github repository https://github.com/isa-ulisboa/greends-hack2023-wildfire

 

Load details

Photo by Egor Vikhrev on Unsplash

List of Hackathon workshops