Refine
Document Type
- Article (1)
- Doctoral Thesis (1)
Language
- English (2)
Has Fulltext
- yes (2)
Keywords
- Customer service (1)
- IceCube Gen2 (1)
- IoT data (1)
- astroparticle physics (1)
- big data (1)
- decision support system (1)
- explainable AI (1)
- industrial AI (1)
- mDOM (1)
- machine learning (1)
Institute
- Department Physik (1)
- Technische Fakultät (1)
The IceCube neutrino observatory is the largest operating neutrino telescope at the moment.
It consists of 5160 Digital Optical Modules (DOMs) on 86 vertical strings buried in a
depth of 1.5km to 2.5km within the Antartic ice instrumenting a volume of approximately
1 cubic km . An upgrade of the in-ice array to a volume of almost 10 cubic km, called IceCube Gen2 high-
energy array, is the subject of current research. The multi photomultiplier (PMT) Digital Optical
Module (mDOM), which consists of 24 symmetrically distributed 3-inch PMTs, is considered
as detection unit for the Gen2 high-energy array. Alternatively, an upgraded version of the
IceCube DOM, called PINGU Digital Optical Module (PDOM), containing only one 10-inch PMT
facing downwards, is also considered as detection unit. This work analyzes the effect of the
sensor segmentation of the mDOM on the angular resolution of through-going muon tracks in
comparison to the angular resolution obtained with the PDOM within the context of a Gen2
high-energy array geometry.
In order to eliminate the effect of different photon detection efficiencies of the two sensor designs,
the quantum efficiencies of the respective PMTs are scaled in the simulation to ensure an
equalized effective photocathode area per module. For down-going and horizontal through-going
muons with an energy between 3TeV and 70PeV a detector equipped with mDOMs yields between
10% and 40% better angular resolution in almost all energy regimes after sensor-independent
quality cuts (based on Monte Carlo information) have been applied.
For up-going muons with energies below 1PeV the upscaled PDOM yields between 7% and 13% lower angular errors.
Finally, estimations of the 90% exclusion limits and the 5σ discovery fluxes of neutrino point
sources are conducted for both sensors in a Gen2 high-energy array.
For sources with a declination below 5° the upper limits and discovery fluxes obtained with the mDOM are 8 − 11% lower.
The upscaled PDOM leads to 4 − 12% lower exclusion limits and discovery fluxes for sources with a declination above 33°.
The most important goal of customer service is to keep the customer satisfied. However, service resources are always limited and must prioritize specific customers. Therefore, it is essential to identify customers who potentially become unsatisfied and might lead to escalations . Data science on IoT data (especially log data) for machine health monitoring and analytics on enterprise data for customer relationship management (CRM) have mainly been researched and applied independently. This paper presents a data-driven decision support system framework that combines IoT and enterprise data to model customer sentiment and predicts escalations. The proposed framework includes a fully automated and interpretable machine learning pipeline using state-of-the-art methods. The framework is applied in a real-world case study with a major medical device manufacturer providing data from a fleet of thousands of high-end medical devices. An anonymized version of this industrial benchmark is released for the research community based on the presented case study, which has interesting and challenging properties. In our extensive experiments, we achieve a Recall@50 of 50.0 % for the task of predicting customer escalations. In addition, we show that combining IoT and enterprise data can improve prediction results and ease troubleshooting. Additionally, we propose a practical workflow for end-users when applying the proposed framework.