Veebipõhine tööriist ennustuspõhiseks protsessianalüütikaks
Files
Date
2017
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Äripotsesside ennustav seire kasutab äriprotsesse toetavaid sündmuste logisid, luues ennustusi äriindikaatorite suhtes, näiteks aeg, maksumus ja järgnevad sündmused. Arvestades kasvavat huvi valdkonna vastu, on nii äri- kui akadeemilises valdkonnas loodud mitmeid lahendusi visualiseerimaks neid ennustusi kasutaja jaoks arusaadaval kujul. Käesolevas magistritöös pakume välja veebipõhise raamistiku ja tööriista, mis lubab kasutajatel hindamiseks kiirelt visualiseerida ennustusmudeleid. Veelgi enam, selle töö käigus loodi veebirakendus, mis suudab suhelda veebiteenuste abil ükskõik millise ennustusprotsessis kasutatava serveri rakendusega. Lõpuks uurib antud töö ka mitme juhtumi põhist ennustust, mis tähendab, et mudelid luuakse kasutades mitut sündmuste logi juhtumit.
Predictive Process Monitoring aims at exploiting event logs in business processes by providing predictions and forecasts on key business metrics such as time, cost and activity executions. As the interest in this field grows, various methods and approaches have been implemented in both academia and industry sectors in order to produce visual results that are understandable to the users. In this Master's Thesis, we propose a web-based framework and tool that enables participants in this field to build quick visualizations on their predictive models for evaluation. Furthermore, this project intends to have an independent front-end application which can work with any method running on the back-end as a web-service that is used in the prediction process. Finally, this project looks into the realm of inter-case predictions which uses multiple cases in building a prediction model of an event log.
Predictive Process Monitoring aims at exploiting event logs in business processes by providing predictions and forecasts on key business metrics such as time, cost and activity executions. As the interest in this field grows, various methods and approaches have been implemented in both academia and industry sectors in order to produce visual results that are understandable to the users. In this Master's Thesis, we propose a web-based framework and tool that enables participants in this field to build quick visualizations on their predictive models for evaluation. Furthermore, this project intends to have an independent front-end application which can work with any method running on the back-end as a web-service that is used in the prediction process. Finally, this project looks into the realm of inter-case predictions which uses multiple cases in building a prediction model of an event log.