Service Analytics

Permanent URI for this collection


Recent Submissions

Now showing 1 - 6 of 6
  • Item
    Feeding-Back Error Patterns to Stimulate Self-Reflection versus Automated Debiasing of Judgments
    ( 2023-01-03) Balla, Nathalie ; Setzer, Thomas ; Schulz, Felix
    Automated debiasing, referring to automatic statistical correction of human estimations, can improve accuracy, whereby benefits are limited by cases where experts derive accurate judgments but are then falsely "corrected". We present ongoing work on a feedback-based decision support system that learns a statistical model for correcting identified error patterns observed on judgments of an expert. The model is then mirrored to the expert as feedback to stimulate self-reflection and selective adjustment of further judgments instead of using it for auto-debiasing. Our assumption is that experts are capable to incorporate the feedback wisely when making another judgment to reduce overall error levels and mitigate this false-correction problem. To test the assumption, we present the design and results of a pilot-experiment conducted. Results indicate that subjects indeed use the feedback wisely and selectively to improve their judgments and overall accuracy.
  • Item
    Context-based Pricing for Revenue Optimization with Applications to the Airline Industry
    ( 2023-01-03) Ettl, Markus ; Subramanian, Shiva ; Drissi, Youssef ; Sun, Wei ; Biggs, Max
    Most airlines use dynamic pricing to optimize the price of their base economy product by maximizing the expected revenue. However, when it comes to pricing of premium products, airlines often uses a static price increments that are applied to the best available economy fare based on simple business rules for adjusting the price based on supply. In this paper, we present a suite of machine learning algorithms that take advantage of the rich booking session context available at the time of the booking to make its predictions. The challenge is to accurately predict bookings for new combinations of attributes by market and segment (departure time, length of stay, advance purchase, length of haul, …) while accounting for cross-product price effects in a scalable manner. To generate practical pricing policies, the approach accommodates a variety of real-world business requirements into the decision optimization problem. We present a scalable approach based on a novel path-based mixed-integer program (MIP) reformulation that can efficiently recover near-optimal pricing policies. We demonstrate the efficacy of our model with extensive experiments on synthetic and real-life data. Finally, we present an airline case study on deriving profitable prescriptive policies for premium cabin tickets based on easily interpretable pricing rules.
  • Item
    Towards More Convenient Services: A Text Analytics Approach to Understanding Service Inconveniences in Digital Platforms
    ( 2023-01-03) Amat-Lefort, Natalia ; Barnes, Stuart
    In today’s fast-paced world, where time is our most valuable asset, convenience is on the rise. This trend has led to a huge growth in digital on-demand services, which target convenience-oriented consumers. Using big data and text analytics, we examine the impact of service inconveniences on customer satisfaction in the context of on-demand food delivery. Building on the Model of Service Convenience and Attribution Theory, we analyze 235,147 user-generated reviews through a combination of keyword-assisted topic modelling and cumulative link model analysis. We introduce the concept of Remote support inconvenience and identify the key topics related to each inconvenience. We find that all service inconveniences negatively influence satisfaction (especially cancelled orders and remote support incidences), and the effects are exacerbated when moderated by stability or controllability attributions. These insights contribute to our theoretical understanding of service inconvenience and can help platforms identify and improve critical areas of their services.
  • Item
    How Artificial Intelligence Can Help the Prediction of Treatment Outcomes of Tuberculosis: A Systematic Literature Review
    ( 2023-01-03) Lino Ferreira Da Silva Barros, Maicon Herverton ; Da Silva Neto, Sebastião Rogerio ; Almeida Rodrigues, Maria Gabriela ; De Souza Sampaio, Vanderson ; Endo, Patricia Takako
    Tuberculosis (TB) is a disease with a global impact that over the years has mainly affected the poorest countries. After confirming the TB diagnosis, the health professional needs to analyze the severity of the clinical situation of the patient in order to make decisions about their treatment, which may include admission to Intensive Care Unit (ICU). The aim of this paper is to present a systematic review focused on Machine Learning (ML) models for predicting TB treatment outcomes. From 253 articles found through a boolean search, only 12 of them were classified as relevant, presented and discussed in this work. Results show that the current literature is focused on binary classification, mainly using tree-based ML algorithms. Based on the results of this systematic review, we state that there are many opportunities to develop new scientific projects in this area, highlighting the need for rigorous methodology to conduct models' configuration as well as experiments to evaluate them.
  • Item
    Introduction to the Minitrack on Service Analytics
    ( 2023-01-03) Fromm, Hansjoerg ; Kühl, Niklas ; Satzger, Gerhard ; Setzer, Thomas
  • Item
    PULL: Reactive Log Anomaly Detection Based On Iterative PU Learning
    ( 2023-01-03) Wittkopp, Thorsten ; Scheinert, Dominik ; Wiesner, Philipp ; Acker, Alexander ; Kao, Odej
    Due to the complexity of modern IT services, failures can be manifold, occur at any stage, and are hard to detect. For this reason, anomaly detection applied to monitoring data such as logs allows gaining relevant insights to improve IT services steadily and eradicate failures. However, existing anomaly detection methods that provide high accuracy often rely on labeled training data, which are time-consuming to obtain in practice. Therefore, we propose PULL, an iterative log analysis method for reactive anomaly detection based on estimated failure time windows provided by monitoring systems instead of labeled data. Our attention-based model uses a novel objective function for weak supervision deep learning that accounts for imbalanced data and applies an iterative learning strategy for positive and unknown samples (PU learning) to identify anomalous logs. Our evaluation shows that PULL consistently outperforms ten benchmark baselines across three different datasets and detects anomalous log messages with an F1-score of more than 0.99 even within imprecise failure time windows.