Loading...
Publication
O Poder da Informação no Planeamento da Distribuição
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 8.53 MB | Adobe PDF |
Authors
Advisor(s)
Abstract(s)
Atualmente, os profissionais da cadeia de abastecimento (CA) enfrentam uma sobrecarga de dados relacionados com as operações logísticas, o que incentiva a adoção de técnicas avançadas de análise de dados para otimizar processos e melhorar a competitividade. Compreender a forma, de como as empresas utilizam o Big Data Analytics (BDA) para aperfeiçoar os processos de distribuição e a sua influência nas decisões estratégicas é fundamental para aumentar a eficiência e a competitividade no mercado.
Este estudo investiga, a forma, de como o Big Data Analytics (BDA) pode ser utilizado para aperfeiçoar os processos de distribuição, com foco na previsão da procura e no suporte à tomada de decisão estratégica, definindo dois objetivos principais. O primeiro consiste na realização de uma análise abrangente da literatura sobre a aplicação de tecnologias de BDA na Gestão da Cadeia de Abastecimento (GCA). Através de uma análise bibliométrica, foram mapeadas as tendências emergentes e identificados os temas mais relevantes, oferecendo uma perspetiva clara sobre o contexto atual e o futuro da logística. O segundo, motivado pela necessidade prática do operador logístico, é desenvolver um modelo de previsão que antecipe a quantidade de pedidos de transporte, para as regiões em estudo, e que permita auxiliar o operador na otimização das suas operações nacionais, contribuindo para a melhoria da sua eficiência logística, promovendo a utilização mais inteligente dos seus recursos e aumentando a sua capacidade de resposta em áreas estratégicas.
Foi adotada uma metodologia quantitativa, focada na análise de séries temporais dos dados históricos de um operador logístico, utilizando diferentes modelos preditivos (Seasonal AutoRegressive Integrated Moving Average (SARIMA), PROPHET, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) e Multi-Layer Perceptron (MLP)). Cada modelo foi testado para prever a quantidade de pedidos de transporte diários. De forma a comparar a eficácia dos modelos em termos de precisão, foram utilizadas métricas de avaliação (Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) e Symmetric Mean Absolute Percentage Error (SMAPE)), que permitiram medir o desempenho de cada um.
Os modelos SARIMA e PROPHET apresentaram um desempenho superior, captando de forma mais eficaz os padrões sazonais, enquanto os modelos baseados em redes neurais, como o LSTM e o GRU, revelaram-se promissores na identificação de padrões não lineares e variações mais complexas nas séries temporais, embora tenham sido identificadas limitações significativas nos modelos.
O estudo acaba por destacar o papel central da análise preditiva na logística moderna, destacando a sua importância para a tomada de decisão baseada em dados e oferecendo uma contribuição relevante para o setor, alinhando a teoria com a prática operacional.
Currently, supply chain (CA) professionals face an overload of data related to logistics operations, which encourages the adoption of advanced data analysis techniques to optimize processes and improve competitiveness. Understanding how companies use Big Data Analytics (BDA) to improve distribution processes and its influence on strategic decisions is fundamental to increasing efficiency and competitiveness in the market. This study investigates how Big Data Analytics (BDA) can be used to improve distribution processes, focusing on demand forecasting and supporting strategic decision making, defining two main objectives. The first consists of carrying out a comprehensive analysis of the literature on the application of BDA technologies in Supply Chain Management (GCA). Through a bibliometric analysis, emerging trends were mapped, and the most relevant themes were identified, offering a clear perspective on the current context and future of logistics. The second, motivated by the practical need of the logistics operator, is to develop a forecast model that anticipates the number of transport requests for the regions under study, and that allows the operator to assist in optimizing its national operations, contributing to the improvement of its logistical efficiency, promoting the smarter use of its resources and increasing its responsiveness in strategic areas. A quantitative methodology was adopted, focused on the analysis of time series of historical data from a logistics operator, using different predictive models (Seasonal AutoRegressive Integrated Moving Average (SARIMA), PROPHET, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) and Multi-Layer Perceptron (MLP)). Each model was tested to predict the number of daily transportation requests. To compare the effectiveness of the models in terms of accuracy, evaluation metrics were used (Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) and Symmetric Mean Absolute Percentage Error (SMAPE)), that allowed each person's performance to be measured. The SARIMA and PROPHET models showed superior performance, capturing seasonal patterns more effectively, while models based on neural networks, such as LSTM and GRU, showed promise in identifying non-linear patterns and more complex variations in time series, although significant limitations have been identified in the models. The study ends up highlighting the central role of predictive analysis in modern logistics, highlighting its importance for data-based decision making and offering a relevant contribution to the sector, aligning theory with operational practice.
Currently, supply chain (CA) professionals face an overload of data related to logistics operations, which encourages the adoption of advanced data analysis techniques to optimize processes and improve competitiveness. Understanding how companies use Big Data Analytics (BDA) to improve distribution processes and its influence on strategic decisions is fundamental to increasing efficiency and competitiveness in the market. This study investigates how Big Data Analytics (BDA) can be used to improve distribution processes, focusing on demand forecasting and supporting strategic decision making, defining two main objectives. The first consists of carrying out a comprehensive analysis of the literature on the application of BDA technologies in Supply Chain Management (GCA). Through a bibliometric analysis, emerging trends were mapped, and the most relevant themes were identified, offering a clear perspective on the current context and future of logistics. The second, motivated by the practical need of the logistics operator, is to develop a forecast model that anticipates the number of transport requests for the regions under study, and that allows the operator to assist in optimizing its national operations, contributing to the improvement of its logistical efficiency, promoting the smarter use of its resources and increasing its responsiveness in strategic areas. A quantitative methodology was adopted, focused on the analysis of time series of historical data from a logistics operator, using different predictive models (Seasonal AutoRegressive Integrated Moving Average (SARIMA), PROPHET, Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU) and Multi-Layer Perceptron (MLP)). Each model was tested to predict the number of daily transportation requests. To compare the effectiveness of the models in terms of accuracy, evaluation metrics were used (Mean Squared Error (MSE), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE) and Symmetric Mean Absolute Percentage Error (SMAPE)), that allowed each person's performance to be measured. The SARIMA and PROPHET models showed superior performance, capturing seasonal patterns more effectively, while models based on neural networks, such as LSTM and GRU, showed promise in identifying non-linear patterns and more complex variations in time series, although significant limitations have been identified in the models. The study ends up highlighting the central role of predictive analysis in modern logistics, highlighting its importance for data-based decision making and offering a relevant contribution to the sector, aligning theory with operational practice.
Description
Keywords
Big Data Analytics Cadeia de Abastecimento Modelos de Previsão Machine Learning Tomada de Decisão