Browsing by Author "Pinheiro, Paulo"
Now showing 1 - 10 of 13
Results Per Page
Sort Options
- An actionable knowledge discovery system in regular sports servicesPublication . Pinheiro, Paulo; Cavique, LuísThis work presents an actionable knowledge discovery system for real user needs with three steps. In the first step, it extracts and transforms existing data in the databases of the ERP and CRM systems of the sports facilities and loads them into a Data Warehouse. In a second phase, predictive models are applied to identify profiles more susceptible to abandonment. Finally, in the third phase, based on the previous models, experimental planning is carried out, with test and control groups, in order to find concrete actions for customer retention.
- A bi‐objective procedure to deliver actionable knowledge in sport servicesPublication . Pinheiro, Paulo; Cavique, LuísThe increase in retention of customer in gyms and health clubs is nowadays a challenge that requires concrete and personalized actions. Traditional data mining studies focused essentially on predictive analytics, neglecting the business domain. This work presents an actionable knowledge discovery system which uses the following pipeline (data collection, predictive model, retention interventions). In the first step, it extracts and transforms existing real data from databases of the sports facilities. In a second step, predictive models are applied to identify user profiles more susceptible to dropout, where actionable withdrawal rules are based on actionable attributes. Finally, in the third step, based on the previous actionable knowledge some of the values of the actionable attributes should be changed in order to increase retention. Simulation of scenarios is carried out, with test and control groups, where business utility and associate cost are measured. This document presents a bi-objective study in order to choose the more efficient scenarios.
- Data science maturity model: from raw data to pearl’s causality hierarchyPublication . Cavique, Luís; Pinheiro, Paulo; Mendes, Armando B.Data maturity models are an important and current topic since they allow organizations to plan their medium and long-term goals. However, most maturity models do not follow what is done in digital technologies regarding experimentation. Data Science appears in the literature related to Business Intelligence (BI) and Business Analytics (BA). This work presents a new data science maturity model that combines previous ones with the emerging Business Experimentation (BE) and causality concepts. In this work, each level is identified with a specific function. For each level, the techniques are introduced and associated with meaningful wh-questions.We demonstrate the maturity model by presenting two case studies.
- Determinação de padrões de desistência em ginásiosPublication . Pinheiro, Paulo; Cavique, LuísO mercado do Fitness regista uma taxa de abandono na ordem dos 40 a 45%, taxa que é considerada muito elevada, tendo em consideração outros sectores de mercado. Este trabalho tem por principal objetivo, através de técnicas de Classificação, determinar o algoritmo mais adequado para encontrar padrões que permitam prever quais os utentes que irão abandonar ou cancelar a sua inscrição num próximo período, atendendo ao padrão de perfil dos utentes que desistiram nos últimos meses, ao seu padrão de consumo e utilização e a eventuais dados demográficos que se venham a considerar relevantes para o estudo em questão.
- Determinação de padrões de desistência em ginásiosPublication . Pinheiro, Paulo; Cavique, LuísO problema da retenção e das causas que levam à desistência da frequência do ginásio é uma questão que os ginásios e academias de fitness tentam há muito entender e consequentemente evitar. Tendo em atenção que a indústria do Fitness tem apostado na instalação massificada de sistemas avançados de Customer Relationship Management (CRM) existem atualmente bases de dados com dados históricos de grande valia. Este projeto tem por objetivo aplicar técnicas de classificação a estas bases de dados de forma a encontrar um método adequado para determinar padrões que permitam prever quais os utentes que irão abandonar ou cancelar a sua inscrição num período próximo, atendendo ao padrão do perfil de comportamento dos utentes que desistiram nos últimos meses, e criar uma ferramenta que forneça informação que permita aos gestores desses ginásios tomar medidas que permitam prolongar a duração da frequência dos utentes.
- Extracting actionable knowledge to increase business utility in sport servicesPublication . Pinheiro, Paulo; Cavique, LuísThe increase in retention of customer in gyms and health clubs is nowadays a challenge that requires concrete and personalized actions. Traditional data mining studies focused essentially on predictive analytics, neglecting the business domain. This work presents an actionable knowledge discovery system which uses the following pipeline (data collection, predictive model, loyalty actions). In the first step, it extracts and transforms existing real data from databases of the sports facilities. In a second step, predictive models are applied to identify user profiles more susceptible to dropout. Actionable rules are generated based on actionable attributes that should be avoided, in order to increase retention. Finally, in the third step, based on the previous actionable knowledge, experimental planning is carried out, with test and control groups, in order to find the best loyalty actions for customer retention. This document presents a simulation and the measure of the business utility of an actions sequence to avoid dropout.
- Large language model for querying databases in PortuguesePublication . Figueiredo, Lourenço; Pinheiro, Paulo; Cavique, Luís; Marques, NunoThis study introduces a system that helps non-expert users find information easily without knowing database languages or asking technicians for help. A specific domain is explored, focusing on a subscription-based sports facility, which serves as an open-source version of a real case study. Utilizing the star schema, the available data in the database is structured to provide accessibility through Portuguese Natural Language queries. Using a Large Language Model (LLM), SQL queries are generated based on the question and the provided star schema. We created a dataset with 115 highly challenging questions drawn from real-world usage scenarios to validate the correctness of the system. Challenges found during testing, like attribute value interpretation, out-of-scope questions, and temporal interval adequacy issues, highlight the insufficiency of the star schema alone in providing the needed context for generating accurate SQL queries by the LLM. Addressing these challenges through enhanced contextual information shows significant improvement in query correctness, with validation results increasing from 57.76% to 88.79%. This study shows the potential and limitations of LLMs in generating SQL queries from Portuguese Natural Language queries.
- A machine learning framework for uplift modeling through customer segmentationPublication . Pinheiro, Paulo; Cavique, LuísIn uplift modeling, the goal is to identify high-value customers based on persuadable customers, those who make a purchase only if contacted. To achieve this, uplift modeling combines machine learning techniques with causal inference, allowing businesses to refine their customer targeting strategies and focus efforts where they are most profitable. This study proposes a practical and reproducible two-phase procedure for identifying highvalue customers. In the first phase, customers are segmented using decision trees, which offer a transparent and data-driven approach to grouping individuals with similar characteristics. This segmentation lays the groundwork for a meaningful interpretation of customer behavior. In the second phase, uplift is calculated for each customer segment by comparing the outcomes of the treatment and control groups. This enables the identification of customer groups with the highest uplift. A real-world use case further illustrates the value and applicability of the proposed method. To validate model performance, the procedure employs established metrics such as the Qini index and Cohen’s kappa, which provide insights into both the effectiveness and reliability of the uplift estimates. This work presents a decoupled procedure for uplift modeling that leverages well-established libraries, fostering transparency and a clear understanding of the analytical process. A key contribution to uplift modeling and causal inference is the use of decision trees for stratification, which enables the creation of meaningful segments and their evaluation through the average treatment effect. By integrating theory with practical implementation, this work offers a comprehensive framework for uplift modeling that bridges academic rigor and business usability.
- Modelos para incremento da retenção em serviços desportivos regulares : análise preditiva e ações de fidelizaçãoPublication . Pinheiro, Paulo; Cavique, LuísAtrair um novo cliente pode custar até cinco vezes mais do que manter um cliente atual satisfeito e por isso a capacidade de detetar, o mais cedo possível, quais os clientes que irão abandonar ou deixar de adquirir um determinado produto ou serviço, bem como as medidas que podem ser implementadas para evitar esse abandono ou quebra nas vendas são questões que todas as empresas gostariam de ver respondidas. Neste trabalho é apresentado um sistema inteligente que gera conhecimento acionável (“actionable knowledge”) baseado em dados reais e orientado para acções de fidelização de clientes de serviços desportivos regulares onde ocorrem elevadas taxas de cancelamento. Seguindo os passos do Database Marketing o sistema evolui em três fases: numa primeira fase, parte de dados obtidos nas bases de dados dos sistemas ERP e CRM existentes nas instalações desportivas, dos quais extrai, transforma e carrega dados num Data Warehousing; numa segunda fase são aplicados modelos preditivos para identificar perfis mais suscetíveis de abandono; e por fim, são aplicadas ações de fidelização direcionadas a cada um dos perfis encontrados com o objetivo de aumentar a fidelização e a taxa de retenção.
- Modelos para incremento da retenção em serviços desportivos regulares: análise preditiva e ações de fidelizaçãoPublication . Pinheiro, Paulo; Cavique, LuísAs instalações desportivas que oferecem serviços desportivos regulares têm vindo a adotar sistemas ERP e CRM, existindo atualmente bases de dados com dados históricos de grande valia. Neste trabalho demonstramos que aplicando modelos preditivos a estes dados é possível identificar perfis de abandono. Com base nos perfis encontrados é realizado um planeamento de experiências, com grupos de teste e controlo, com vista a encontrar ações concretas de fidelização.