Remaining Time Estimation in Business Processes Using Traces' Structural Information
In this Ph.D. we present a framework for predicting the remaining time of a business process. Our framework consists of building an Extended Annotated Transition System (EATS) model which extends the baseline Annotated Transition System considering eight structural features of the traces, where each...
| Autor: | |
|---|---|
| Tipo de documento: | tese |
| Data de publicação: | 2020 |
| País: | España |
| Recursos: | Universidad de Santiago de Compostela (USC) |
| Repositório: | Minerva. Repositorio Institucional de la Universidad de Santiago de Compostela |
| Idioma: | inglês |
| OAI Identifier: | oai:minerva.usc.gal:10347/23283 |
| Acesso em linha: | http://hdl.handle.net/10347/23283 |
| Access Level: | Acceso aberto |
| Palavra-chave: | 1203.04 Inteligencia Artificial 1203.08 Código y Sistemas de Codificación |
| Resumo: | In this Ph.D. we present a framework for predicting the remaining time of a business process. Our framework consists of building an Extended Annotated Transition System (EATS) model which extends the baseline Annotated Transition System considering eight structural features of the traces, where each state in the EATS is annotated with a partitioned list of attributes of these features. Linear regression is applied to each partition to predict the remaining time. Experimental validation of our model has been conducted with ten real-life benchmark datasets, confronting our estimations to the state of the art. Results show that our model not only outperforms the baseline but also other approaches in the literature. We have also addressed the scalability of our model, by introducing two attribute selection methods which allow us to keep a good balance between the computational cost and acceptable prediction accuracy. |
|---|