La tesi ha l’obiettivo di analizzare il cambiamento organizzativo in una prospettiva di social network analysis, sfruttando dati longitudinali raccolti a seguito della modifica della struttura organizzativa in un Policlinico Universitario italiano. Il manoscritto è organizzativo in tre paper. Il primo paper si focalizza sul tema del rapporto tra network formali e network informali, analizzando come la modifica del primo comporti una corrispondente variazione nel secondo. Il paper dimostra come, in assenza di strutture organizzative ben formalizzate, gli individui tendono ad allacciare nuovi legami con colleghi che appartengono alla stessa specializzazione. Il secondo paper, invece, attingendo prettamente alla letteratura di comportamento organizzativo, analizza il tema della dinamicità del network, fornendo evidenze in relazione alla stabilità del network stesso a seguito del cambiamento. Particolare attenzione, è inoltre, dedicata alle dinamiche intra – team e al ruolo di quest’ultime nell’accettazione o meno del cambiamento. Infine, il terzo paper sviluppa il tema della network density e di come quest’ultima possa essere correlato al cambiamento organizzativo, in termini di reazione al cambiamento. Inoltre, si dimostra come la formalizzazione abbia un impatto positivo sulla densità del network, specie in contesti organizzativi caratterizzati da una bassa gerarchia e coordinamento orizzontale. / This thesis aims to analyze organizational change in a social network analysis perspective, exploiting longitudinal data collected after a modification of the organizational structure in an Italian Teaching Hospital The manuscript is organized into three papers. The first paper focuses on the theme of the relationship between formal networks and informal networks, analyzing how the modification of the first involves a corresponding variation in the second. The paper demonstrates how, in the absence of formalized organizational structures, individuals tend to establish new ties with colleagues who belong to the same specialization. The second paper, drawing purely from the organizational behavior literature, analyzes the issue of the network dynamics , providing evidence and antecedents for network stability in response to organizational change. Particular attention is also given to the intra - team dynamics and the impact of individual perception of collective properties in driving employees in accepting or not the organizational change. Finally, the third paper develops the theme of network density and how the latter can be related to organizational change, in terms of reaction to change. Furthermore, it is shown how formalization has a positive impact on network density, especially in organizational contexts characterized by a low hierarchy and horizontal coordination.

Le routage dans Internet est basé sur des tables dites de routage, formées de blocs d’adresses IP. Cependant, la construction et la maintenance de telles tables de routage nécessitent l’utilisation de protocoles complexes qui ne passent pas à l’échelle en termes de mémoire et d’utilisation CPU. De plus, l’expérience montre que le plan d’adressage IP est insuffisant, car la sémantique d’une adresse IP est à la fois un identificateur et un localisateur. Dans nos travaux, nous proposons un système de réseau recouvrant pair-à-pair libre de toute contrainte topologique et utilisant des coordonnées virtuelles prises dans le plan hyperbolique nommé CLOAK (Covering Layer Of Abstract Knowledge en anglais). Les schémas de routages locaux basés sur des coordonnées virtuelles extraites du plan hyperbolique ont suscité un intérêt considérable ces dernières années. Dans cette thèse, nous proposons une nouvelle approche pour saisir le potentiel de la géométrie hyperbolique. L’objectif est de construire un système extensible et fiable pour créer et gérer des réseaux recouvrants dans Internet. Le système est implémenté comme une infrastructure pair-à-pair structuré basé sur les protocoles de la couche transport entre les pairs. Quant à l’organisation des données dans l’espace virtuel, nous employons la réplication pour améliorer la disponibilité et l’accessibilité des objets de l’overlay potentiellement instable. Nous avons implémenté et évalué différentes méthodes de réplication (réplication radiale, réplication circulaire).A l’aide de simulations, nous évaluons notre proposition à travers un certain nombre de métriques et nous montrons que les réseaux recouvrants pair-à-pair basés sur la géométrie hyperbolique ont de bonnes performances par rapport aux autres DHT existantes tout en introduisant flexibilité et robustesse dans les réseaux recouvrants dynamiques. / Internet routing is based on forwarding tables populated by blocks of IP addresses. However, the construction and maintenance of such tables require the use of complex routing protocols that are typically not scalable in terms of memory and CPU usage. Moreover, experience shows that the IP addressing plane is insufficient due to the semantic of an IPaddress being both an identifier and a locator. In this paper, we propose a P2P overlay system of freed topology and using virtual coordinates taken from the hyperbolic plane named CLOAK(Covering Layer Of Abstract Knowledge en anglais). Local knowledge routing schemes based on virtual coordinates taken from the hyperbolic plane have attracted considerable interest in recent years. In this thesis we propose a new approach for seizing the power of the hyperbolic geometry. We aim at building a scalable and reliable system for creating and managing overlay networks over the Internet. The system is implemented as a structured peer-to-peer infrastructure based on the transport layer connections between the peers. Concerning data organisation in the virtual space, we use replication strategy for improve overlay objects disponibilty and accessibility in context potentially unstable. We have implemented and evaluated various replication methods (radial replication, circular replication). Using simulations, we assess our proposal across a certain number of metric and show that overlay Peer-to-Peer network based on the hyperbolic geometry have good performances in comparison with other existent DHT while introducing suppleness and robustness in the dynamic overlay network.

Submitted by Kelson Anthony de Menezes (kelson@ucb.br) on 2016-10-28T18:48:01Z No. of bitstreams: 1 JeffersonJoseCeruttiGauerDissertacao2016.pdf: 1448138 bytes, checksum: 7c0985d46840a27fe872a0de79761029 (MD5) / Made available in DSpace on 2016-10-28T18:48:01Z (GMT). No. of bitstreams: 1 JeffersonJoseCeruttiGauerDissertacao2016.pdf: 1448138 bytes, checksum: 7c0985d46840a27fe872a0de79761029 (MD5) Previous issue date: 2016-03-10 / Similarities of products and services, market stagnation, the portability of operations among institutions, competition and competitiveness in banking sector have motivated more attention to customer loyalty. It is essential to win new clients as well as to retain them in order to avoid churn. So management tools that concern relations with customers require an increasing amount of variables. Present study covers the best clients in a big-size Brazilian financial institution. It proposes a model for churn predicting, based on the evolution of their loans and investments. Operations from ca. 291 thousands clients were the input data for software QlikView (a user-oriented Business Intelligence platform). The model transformed the Daily Balance Average into a logarithm scale in order to assess the value oscillation according to periods. The achieved index seems to be a possible churn predictor, which indicates that relations management should regard carefully customers susceptible to churn. Nevertheless this index alone does not explain the churn rate. It is recommended to apply it as a complement and a refinement of other indexes that are already deployed in customer loyalty management. / As semelhan??as de produtos e servi??os, a estagna????o do mercado, a portabilidade de opera????es entre institui????es e a concorr??ncia e competitividade no setor banc??rio t??m motivado mais aten????o ?? fideliza????o do cliente. A considera????o de tais fatores ?? essencial para a conquista de novos clientes, bem como para a sua reten????o, a fim de evitar o churn. Assim, ferramentas de gest??o de relacionamento com o cliente exigem uma quantidade crescente de vari??veis. O presente estudo abrange os melhores clientes de uma institui????o financeira brasileira de grande porte. Prop??e um modelo para a predi????o de churn, com base na evolu????o dos seus empr??stimos e investimentos. Opera????es de 291.761 clientes foram os dados de entrada para a ferramenta QlikView (uma plataforma de BI ??? Business Intelligence ??? orientada ao usu??rio). O modelo transformou a M??dia de Saldos Di??rio (MSD) em uma escala logar??tmica, a fim de avaliar a oscila????o de acordo com os per??odos. O indicador alcan??ado parece ser um poss??vel preditor de churn, o que indica que a gest??o de relacionamento deve considerar cuidadosamente os clientes suscet??veis ?? evas??o. No entanto, s?? este indicador n??o explica a taxa de churn. Recomenda-se aplic??-lo como um complemento e um refinamento de outros indicadores que j?? est??o implantados na gest??o da fideliza????o com o cliente.

Artificial neural networks are fascinating machine learning algorithms. They used to be considered unreliable and computationally very expensive. Now it is known that modern neural networks can be quite useful, but their computational expensiveness unfortunately remains. Statistical boosting is considered to be one of the most important machine learning ideas. It is based on an ensemble of weak models that together create a powerful learning system. The goal of this thesis is the comparison of these machine learning models on three use cases. The first use case deals with modeling the probability of burglary in the city of Chicago. The second use case is the typical example of customer churn prediction in telecommunication industry and the last use case is related to the problematic of the computer vision. The second goal of this thesis is to introduce an open-source machine learning platform called H2O. It includes, among other things, an interface for R and it is designed to run in standalone mode or on Hadoop. The thesis also includes the introduction into an open-source software library Apache Hadoop that allows for distributed processing of big data. Concretely into its open-source distribution Hortonworks Data Platform.

This thesis deals with a complete analysis of real data in free to play multiplayer games. The analysis is based on the methodology CRISP-DM using GUHA method and system LISp-Miner. The goal is defining player churn in pool from Geewa ltd.. Practical part show the whole process of knowledge discovery in databases from theoretical knowledge concerning player churn, definition of player churn, across data understanding, data extraction, modeling and finally getting results of tasks. In thesis are founded hypothesis depending on various factors of the game.

This thesis is concerned with analysis of the mobile operators in the Czech Republic. Introduction part of thesis is devoted to the explanation of the key indicators which are used by mobile operators to analyze the rate of customer churn, or average revenue per user, or the customer value. Very importat is also a comparison of price consumption baskets in the Czech Republic with the other countries. The situation has changed dramatically offering unlimited tariffs in the already fully saturated market. In the practical part of this thesis I have focused on a detailed analysis of customers data in Vodafone company a I have reviewed the current offer of affordable tariffs and data services. I tried to implement new tariffs based on a real information of all customers - individuals over a period of 3 months. The aim of the thesis is to propose tariffs and services for customers, which would maintain the current customer base or which would increase the base slightly.

In the digital era and advent of artificial intelligence, digital retailing has emerged as a notable shift in commerce. It empowers e-tailers with data-driven insights and predictive models to navigate a variety of challenges, driving informed decision-making and strategic formulation. While predictive models are fundamental for making data-driven decisions, this thesis spotlights binary classifiers as a central focus. These classifiers reveal the complexities of two real-world problems, marked by their particular properties. Specifically, binary decisions are made based on predictions, relying solely on predicted class labels is insufficient because of the variations in classification accuracy. Furthermore, prediction outcomes have different costs associated with making different mistakes, which impacts the utility. To confront these challenges, probabilistic predictions, often unexplored or uncalibrated, is a promising alternative to class labels. Therefore, machine learning modelling and calibration techniques are explored, employing benchmark data sets alongside empirical studies grounded in industrial contexts. These studies analyse predictions and their associated probabilities across diverse data segments and settings. The thesis found, as a proof of concept, that specific algorithms inherently possess calibration while others, with calibrated probabilities, demonstrate reliability. In both cases, the thesis concludes that utilising top predictions with the highest probabilities increases the precision level and minimises the false positives. In addition, adopting well-calibrated probabilities is a powerful alternative to mere class labels. Consequently, by transforming probabilities into reliable confidence values through classification with a rejection option, a pathway emerges wherein confident and reliable predictions take centre stage in decision-making. This enables e-tailers to form distinct strategies based on these predictions and optimise their utility. This thesis highlights the value of calibrated models and probabilistic prediction and emphasises their significance in enhancing decision-making. The findings have practical implications for e-tailers leveraging data-driven decision support. Future research should focus on producing an automated system that prioritises high and well-calibrated probability predictions while discarding others and optimising utilities based on the costs and gains associated with the different prediction outcomes to enhance decision support for e-tailers. / <p>The current thesis is a part of the industrial graduate school in digital retailing (INSiDR) at the University of Borås and funded by the Swedish Knowledge Foundation.</p>

No description available.

Churn refers to the discontinuation of a contract; consequently, customer churn occurs when existing customers stop being customers. Predicting customer churn is a challenging task in customer retention, but with the advancements made in the field of artificial intelligence and machine learning, the feasibility to predict customer churn has increased. Prior studies have demonstrated that machine learning can be utilized to forecast customer churn. The aim of this thesis was to develop and implement a machine learning model to predict customer churn and identify the customer features that have a significant impact on churn. This Study has been conducted in cooperation with the Swedish insurance company Bliwa, who expressed interest in gaining an increased understanding of why customers choose to leave.  Three models, Logistic Regression, Random Forest, and Gradient Boosting, were used and evaluated. Bayesian optimization was used to optimize the models. After obtaining an indication of their predictive performance during evaluation using Cross-Validation, it was concluded that LightGBM provided the best result in terms of PR-AUC, making it the most effective approach for the problem at hand. Subsequently, a SHAP-analysis was carried out to gain insights into which customer features that have an impact on whether or not a customer churn. The outcome of the SHAP-analysis revealed specific customer features that had a significant influence on churn. This knowledge can be utilized to proactively implement measures aimed at reducing the probability of churn. / Att förutsäga kundbortfall är en utmanande uppgift inom kundbehållning, men med de framsteg som gjorts inom artificiell intelligens och maskininlärning har möjligheten att förutsäga kundbortfall ökat. Tidigare studier har visat att maskinlärning kan användas för att prognostisera kundbortfall. Syftet med denna studie var att utveckla och implementera en maskininlärningsmodell för att förutsäga kundbortfall och identifiera kundegenskaper som har en betydande inverkan på varför en kund väljer att lämna eller inte. Denna studie har genomförts i samarbete med det svenska försäkringsbolaget Bliwa, som uttryckte sitt intresse över att få en ökad förståelse för varför kunder väljer att lämna. Tre modeller, Logistisk Regression, Random Forest och Gradient Boosting användes och utvärderades. Bayesiansk optimering användes för att optimera dessa modeller. Efter att ha utvärderat prediktiv noggrannhet i samband med krossvalidering drogs slutsatsen att LightGBM gav det bästa resultatet i termer av PR-AUC och ansågs därför vara den mest effektiva metoden för det aktuella problemet. Därefter genomfördes en SHAP-analys för att ge insikter om vilka kundegenskaper som påverkar varför en kund riskerar, eller inte riskerar att lämna. Resultatet av SHAP-analysen visade att vissa kundegenskaper stack ut och verkade ha en betydande påverkan på kundbortfall. Denna kunskap kan användas för att vidta proaktiva åtgärder för att minska sannolikheten för kundbortfall.

This study investigates the performance of Logistic Regression, k-Nearest Neighbors (KNN), and Random Forest algorithms in predicting customer churn within an e-commerce platform. The choice of the mentioned algorithms was due to the unique characteristics of the dataset and the unique perception and value provided by each algorithm. Iterative models ‘examinations, encompassing preprocessing techniques, feature engineering, and rigorous evaluations, were conducted. Logistic Regression showcased moderate predictive capabilities but lagged in accurately identifying potential churners due to its assumptions of linearity between log odds and predictors. KNN emerged as the most accurate classifier, achieving superior sensitivity and specificity (98.22% and 96.35%, respectively), outperforming other models. Random Forest, with sensitivity and specificity (91.75% and 95.83% respectively) excelled in specificity but slightly lagged in sensitivity. Feature importance analysis highlighted "Tenure" as the most impactful variable for churn prediction. Preprocessing techniques differed in performance across models, emphasizing the importance of tailored preprocessing. The study's findings underscore the significance of continuous model refinement and optimization in addressing complex business challenges like customer churn. The insights serve as a foundation for businesses to implement targeted retention strategies, mitigating customer attrition, and promote growth in e-commerce platforms.