Loading...
PM4PREF – Digital platform for logistics process analysis and tracking with smart packaging, using Process Mining techniques
| Name: | Description: | Size: | Format: | |
|---|---|---|---|---|
| 5.31 MB | Adobe PDF |
Authors
Abstract(s)
Modern logistics operations generate vast amounts of process data but often lack tools
that transform these records into actionable insights. This report addresses this gap by
developing a decision-support platform for historical statistics, real-time monitoring,
predictive analysis, and conformance checking of deliveries. The research is motivated
by challenges faced at Prozis, a major European e-commerce company, where handling
fragile and temperature-sensitive products requires timely detection of risks such as
delays, route deviations, and inadequate environmental conditions.
Following the Design Science Research Methodology (DSRM), the project Decision
Support System – Intelligent Package (DSSIP) was designed and implemented. The
system integrates process mining techniques with Internet of Things (IoT) sensor data
and geolocation streams to provide transparency across ongoing deliveries. Its core
modules include: historical statistics for exploratory analysis; a conformance checker
using GPS based clustering to detect deviations from reference routes; predictive modeling
of activity timestamps based on polynomial regression and contextual filtering;
and a real-time dashboard for monitoring package conditions such as temperature, humidity,
and shocks.
Validation was carried out through experiments on historical event logs and simulated
sensor datasets, combined with real world case studies from Prozis. Results show
that contextualized predictions improve accuracy compared to global models, frequent
transitions yield robust forecasts, and low-support activities expose limitations in process
discovery. Usability testing (PSSUQ overall score: 1.93) confirmed that the system
is effective and well-accepted by users.
This research contributes to the state of the art by incorporating spatio-temporal
data into process mining in a logistics setting and advancing context-aware predictive
monitoring.
The results emphasize the potential of process mining and IoT integration to enhance
logistics resilience, reduce waste, and improve service quality in large-scale ecommerce
operations.
Description
Keywords
Process mining Predictive monitoring Spatio-temporal analysis Usability Testing IoT Data Decision support systems