Publicação
Bridging the Gap: Developing an Online Metrics Calculation Pipeline for Improved Product Learning to Rank in E-commerce
| datacite.subject.fos | Ciências Naturais::Ciências da Computação e da Informação | pt_PT |
| dc.contributor.advisor | António, Nuno Miguel da Conceição | |
| dc.contributor.author | Gomes, Madalena Botelho Silva | |
| dc.date.accessioned | 2024-04-17T11:19:21Z | |
| dc.date.available | 2024-04-17T11:19:21Z | |
| dc.date.issued | 2024-04-16 | |
| dc.description | Project Work presented as the partial requirement for obtaining a Master's degree in Data Science and Advanced Analytics, specialization in Business Analytics | pt_PT |
| dc.description.abstract | This project involved the development and implementation of a Machine Learning pipeline to track product ranking performance metrics at FARFETCH, a leading platform in the luxury e-commerce sector. The aim was for the Data Science team to start monitoring metrics that would allow them to evaluate the model's performance, optimize it, and consequently, improve the user experience in PLPs as well as the company's commercial success. Using a data-driven approach, this project integrates information on user interaction and sales transactions, which allows various types of signals to be obtained, namely clicks, add to bag, add to wishlist, and bought, in order to monitor ranking metrics daily. This study allowed the team to track historical data over time, which facilitated the process of detecting possible problems that could affect the model's performance other than the external factor of seasonality. In a short space of time, the team was able to identify what could be causing a decrease in the model performance, in this case, the transactions table, which was important for training the model, had been altered and was incomplete. Solving the problem resulted in a substantial improvement in the training and prediction metrics, which validated the model's optimization. The creation of the pipeline is fundamental for making informed decisions, detecting patterns and variations in the model's behavior. As such, the results obtained emphasize the benefit of a data-driven approach to optimize ranking algorithms and business strategy in general. Finally, the research highlights the importance of selecting and evaluating metrics to measure the performance of any Machine Learning ranking system, contributing to Data Science applied to e-commerce. | pt_PT |
| dc.identifier.tid | 203586298 | pt_PT |
| dc.identifier.uri | http://hdl.handle.net/10362/166320 | |
| dc.language.iso | eng | pt_PT |
| dc.rights.uri | http://creativecommons.org/licenses/by/4.0/ | pt_PT |
| dc.subject | Machine Learning | pt_PT |
| dc.subject | Data Science | pt_PT |
| dc.subject | Data Engineering | pt_PT |
| dc.subject | Pipeline | pt_PT |
| dc.subject | Product Ranking Analytics | pt_PT |
| dc.subject | E-commerce | pt_PT |
| dc.subject | SDG 8 - Decent work and economic growth | pt_PT |
| dc.title | Bridging the Gap: Developing an Online Metrics Calculation Pipeline for Improved Product Learning to Rank in E-commerce | pt_PT |
| dc.type | master thesis | |
| dspace.entity.type | Publication | |
| rcaap.rights | openAccess | pt_PT |
| rcaap.type | masterThesis | pt_PT |
| thesis.degree.name | Mestrado em Ciência de Dados e Métodos Analíticos Avançados, especialização em Métodos Analíticos para a Gestão | pt_PT |