Summary:
As quantities of data being generated have grown dramatically and decision-making process and knowledge creation have become increasingly more dependent on the said data. Given that a company’s decisions are never better than the quality of its data, the need to overcome complexity
of standardizing business process and improve data quality, data management has become more centralized. This gave rise to the concept of Master Data Management (MDM), which can be defined as “an application-independent process which describes, owns and manages core business
data entities”. The core areas of MDM are processes, IT, data quality and data governance based on literature. This strategic case study of a FMCG firm based in Switzerland and focuses on what the different business processes and practices around MDM are and how they could be improved.
The master data in question are the five main attributes of a case – the main way of registering customer interactions – on an e-commerce platform built and supported by Isobar.
Results:
A number of suboptimal processes were identified based on observations made by the author and comprehensive improvements were proposed and their implementation discussed. Several improvements had to do with reducing information asymmetry via information sharing practices that would decrease ticket resolution time. Other improvements had to do with upload of Master Data in order to decrease total number of clicks and total upload time. Other major improvements are related to request tickets for Master Data generation that aim to improve both data quality and governance. Additionally, several initiatives are proposed that would further improve data quality and enhance user experience.
Summary:
Inventories are roughly divided in two categories: articles which make profit and articles which make losing money. The first ones are the star products generally fast moving with high quantities whereas the second are the slow movers which freeze working capital without contributing much to the profit. To address this inventory issue, several tools were developed across the years, among which the product life cycle management and the product classification. One common purpose of performing such analysis is to identify the potential actions to take on each product typology: increasing the margins for profitable article and reduce operational costs for slow movers. In this thesis, we aim to provide a general framework to carry out a product classification in order to optimize margins and to reduce inventories costs. In a first time, we propose a product life cycle management method based on logistics needs capable of capturing both markets inputs and logistical information. Then we propose a conceptual framework for products clustering in order to optimize inventory management: traditionally, product classifications are based on a single criteria which is the annual value consumed, we first develop an alternative model which propose to build a classification criterion dedicated to the specificities of each dataset. The model is based on a criterion derived from a return on asset maximization process. The new criterion proposition is based on inherent product information and on the dataset analysis, it provides an in-depth analysis and a business tailored classification of product with the objective of maximizing profit while minimizing costs. To illustrate this new criterion proposition, we first execute the proposed methodology with a dataset provided by a watch manufacturing company, then, we compare the results obtained in term of inventory investment and annual running costs with the commonly used criteria to show that it better suits the company’s needs. Based on that criteria comparison, we also confirm results from several authors in the literature, mentioning that the most common criterion (annual value) performance is rather poor. Moreover we present a discussion about our findings on the best number of classes and the classes’ size. To conclude the inventory optimization process, we propose to define an optimal service level to compute safety stocks under an optimal lot sizing policy.
Summary:
Employee rostering is a process of assigning available employees to open shifts. Automating it has ubiquitous practical benefits for nearly all industries such as reducing manual workload and producing flexible, high-quality schedules. In this project, a mixed integer linear program (MILP) was developed to optimize employee rostering for Swissgrid where it is still a largely manual process. To solve larger problems more tractably, a hybrid methodology was also developed which combined the MILP and scatter search, an evolutionary algorithm that has achieved success in many problem domains. The employee rostering model also aptly served as a simulator to evaluate the adequacy of any given number of employees so as to find, ultimately, the optimal staffing level in a company – an important aspect of strategic planning. To this end, golden section search technique was applied to find the optimal number of employees efficiently under a number of different parameter settings.
Results:
On many use cases, the optimal annual roster from solving the MILP with CPLEX (a commercial solver) guaranteed compliance with labor laws, maximized employees’ preference satisfaction, and distributed the workload as equally as possible among them, all the while taking less than 15 minutes on average for a team of 12 employees. The hybrid methodology of CPLEX and scatter search proved to be a highly robust and efficient algorithm, solving many realistic problems of varying complexity from medium (24 employees) to large (72 employees) in a time that was an order of magnitude faster than CPLEX on its own. Lastly, the results from optimizing the number of employees with golden section search suggested that appropriate training to improve each employee’s versatility, such as having more skills to perform other types of duty, could allow Swissgrid to optimize its workforce without necessarily compromising employees’ preference satisfaction.
Summary:
Understanding customer needs is key to the success of every company. However, large companies often struggle to understand the preferences and needs of their customers due to the extreme difficulty of finding those at scale. Attempts to understand customers with broad characteristics like age, gender, and so on often lead to dramatic failures given the diversity and ever-changing customer preferences. Natural language processing and machine learning models can give us both the scalability and the accuracy that is needed for consumer insight generation. In order to train such sophisticated models, we intend to bring various datasets, such as market studies and online reviews. We then use deep learning models and NLP techniques to both help process unstructured texts to combine those market studies together and extract consumer insights by training predictive models.
Results:
We have achieved the following results:
– Retrieving semantically similar documents/sentences. In this tool, we have used word2vec embedding in combination with Word Mover’s Distance for fast retrieval of the semantically closest sentences.
– Intelligently merging similar questions/sentences from different market studies. For this tool, we have used a novel deep learning model that can compare pairs of sentences and merge similar ones.
– A text classification tool that can extract consumer insights from unstructured texts. As a starting point, we have trained our model to predict the sentiment of the comments accurately.
– A web application that can facilitate interacting with these tools, in addition to providing additional tools for performing in-depth analysis.
Summary:
Depuis le 1er janvier 2018, grâce à un cadre légal innovant et avantageux, il est envisageable de partager la production de sources d’énergies renouvelables décentralisées et locales au sein d’un regroupement pour la consommation propre (RCP). Cela signifie qu’il est aujourd’hui possible de créer des micro réseaux communautaires ayant une production locale mutualisée entre ses membres tout en étant raccordés au réseau de distribution conventionnel pour sécuriser l’approvisionnement. Les Energéticiens ont donc saisi cette occasion pour développer de nouveaux modèles d’affaires très attractifs favorisant la transition énergétique et étant résiliants dans le cas d’une ouverture future du marché de l’électricité.
Toutefois, le traitement des données de comptage au sein de ces regroupements nécessite de réinventer les processus standards appliqués par les gestionnaires du réseau de distribution tel que Romande Energie.
Cette étude définit le flux de données utiles pour les RCP et analyse les différentes stratégies et prestations adaptées à Romande Energie.
Results:
Ce Projet de Master pose les bases pour le choix d’une solution optimale concernant le traitement des données compteurs des produits RCP de contracting.
Les principaux résultats sont les suivants :
- Un marché pair-à-pair du kWh régi par la blockchain n’est viable ni financièrement ni d’un point de vue environnemental
- L’état du marché des prestations de décompte et de facturation a été établi
- Deux stratégies concernant la gestion des données de comptage ont pu être définies
- Trois prestataires ont été pré-sélectionnés
Summary:
In this thesis, we address several issues within the global supply chain of a world-renowned Swiss luxury watch company. We focus on the product portfolio segmentation, information flow structure and logistic flows for simplified management and distribution of material watch straps, sourced from two external leather strap suppliers. The overall objective of our project is to reduce the lead times on a global level for customer orders of material straps.
Within the luxury industry, excellent customer satisfaction is of paramount importance. The quality of customer service has a large impact on a company’s image and it is therefore imperative for this prestigious company, as a leader in the luxury watch industry, to excel in this field.
In this thesis, we offer improvements to the current system by simplifying the structure of the portfolio and the internal management of strap orders, resulting in a reduction in supplier lead times, while simultaneously maintaining a high quality as well as offering an unlimited choice of watch straps to the final customer. The topics covered in this thesis are product portfolio segmentation, product variety analysis and simplification, a fuzzy SWOT analysis of possible logistic flows and a flow classification for points of sale.
Results:
By combining both qualitative and quantitative methods, we have developed a promising solution that is estimated to satisfy the demand for “Collection” straps met by 95% of the points of sale within a period of two weeks, where there currently exists an average lead time of six weeks, and “Exclusive” straps within six weeks for the entire global market, where there currently exists a worst case scenario lead time of five months. We propose that “Collection” and “Exclusive” straps should represent 80% and 15% of order volume respectively.
Summary:
Buy and builds are investment strategies whereby a private equity firm builds value by leveraging the superior position of a platform company to make multiple acquisitions and create scale. I develop a model breaking down value creation in buy and builds along the holding period. Notably, I make the case, both theoretically and empirically, that key success factors lie in the early stage of the integration, when the firms come together and collaboratively shape their new organization. Precisely I find that optimizing the degree of integration in the new organization, establishing a strong governance to implement the integration, creating an atmosphere for collaboration, retaining talent and support from the private equity firm are high-level success factors increasing performance in buy and builds.
Results:
I find, at a deeper level, that the retention of key talent and the establishment of a strong governance can be mutually beneficial, as involving the most talented employees in the governance process gives them an incentive to remain in the company while strengthening the governance with knowledgeable individuals.
Furthermore, aligning cultures, communicating transparently and ensuring the alignment of managers and sponsor contribute to creating an atmosphere for collaboration needed for the full realization of synergies. The definition of the TOM can also boost trust and excitement across companies as it allows projecting oneself to the integration desired product.
Additionally, the investment thesis, defined by the sponsor through the company and industry diligence, and transparency on the standalone companies should be inputs to the Target Operating Model discussion.
Ultimately, transfer of skills, resources, culture and people should unify the merged company, which combined with a superior revenue and cost structure and the larger size resulting from the buildup, should give the new organization higher growth prospects in turn contributing to multiple expansion. I expect this contribution to be important as theory suggests multiple expansion is the number one driver of performance in buildups.
The diligence of industry and company in the pre-deal phase should endeavour to uncover key success characteristics of buy and build such as, but not limited to, a fragmented, stable industry, and a platform with a strong management team and scalable sales.
Finally, the investment thesis, and therefore the nature of the invested industry and companies, governs the size and diversification profile of the investment portfolio, from which multiple expansion is expected.
Summary:
The project consists of an analysis of the present process flow of the sorting and assembly of balance wheels in the watchmaking department of a prestigious watch manufacture. The initial study of the present state is followed by a proposition of a new process flow and an improvement of the process layout. Finally, the implementation road map and next steps are listed along with the future challenges.
The objectives of the following work include the reduction of the transit time for the WIP, the elimination of a stock level in the assembly line and the increase in traceability of the semi-finished product. The following improvements will ensure a better flow of goods in the assembly line of balance wheels, a higher efficiency and a better quality control.
A statistical tool is developed in order to guide the purchasing decisions of the procurement team. The tool predicts the potential statistical distribution of components to be received based on the historical data of previous deliveries and determines the number of balance wheels that can be assembled. The computation is done on a rolling horizon in order to account for a change in the supplier’s accuracy and an increase in the production capabilities and competencies. The tool developed is updated automatically on a daily basis to provide up-to-date information to the firm’s management
Results:
The suggested new flow process guarantees the traceability of the balance wheel, the reduction of the volume of balance wheels ordered and the elimination of an intermediate stock level. The proposed merger and line balancing of the sorting and assembly workshop will ensure a more flexible and responsive supply chain. A first pilot program implementation of the newly proposed tool and process on a unique reference will start in the last quarter of 2020 and span over a period of 4 months. If by the end of this pilot test the new process is able to meet the objectives set previously: the decrease of inventory levels, the ability to reach the safety stock levels and to maintain traceability; the tool and new process will be rolled out on all the other references.
Summary:
This project analyzed, from the perspective of project managers and the project management office, the organizational design changes that the Life Sciences department of a global company is going through. The general acceleration of business combined to overall changes in paradigms mean that organizations have to pick-up the pace and adapt to rapidly changing environments. Failure to adapt results in loss of competitive advantage, market share and can even cause companies to go out of business. Transformations are risky and disruptive endeavors and even established, industry leading companies, sometimes fail to reinvent and transform themselves. Organizational design and change must be analyzed comprehensively as it incorporates the company strategy, structure, processes and its people. Due to the context during which this work has taken place, it will also explore the effects of remote work on meeting dynamics as meeting are a central communication process in organizations and are critical to successful change initiatives. Additionally, the potential lasting impact of the current pandemic on the standard work week is explored as it will have to be taken into consideration in the design choices made by the company.
Results:
- Analysis of the organizational design choices and consequences of the Life Sciences department using a well-established framework
- Exploration of if and how the first changes implemented by the organization are beneficial
- Examination of the meetings dynamics and how they evolve in a fully remote situation
- Identification of potential change-related, company specific, critical success factors
- Examination of the effects and potential lasting impact of the COVID-19 pandemic on the future ways of working of the company
Summary:
The main goal of this study was to work on the entire project development process for development of new product to propose improvements. First, the current situation is analyzed and the literature on Agile methods is reviewed. Then, a focus has been made on the Design Sprint method. This agile methodology created at GV (Google’s new technology investment
platform, formerly Google Venture) and inspired by IDEO’s Design Thinking and by the Agile approach used in the Silicon Valley, has the ability to focus the innovation process on the user, while activating a strong collaboration between participants. After adaptation to apply the method at the very beginning of the projects, this 5-day process has been experienced. Finally,
the study capitalized on this experiment and proposed tailored solutions for the company to reduce the timeline and the costs of the projects.
Results:
- Creation of the version 2 of the project development process. On this version the links between actions and the responsibilities appear clearly. It facilitates the immersion and learning of this flow and allow people who join and will join the company to be operational and proactive very quickly. It also allows to obtain a more realistic estimate of the timeline for each project.
- Adaptation of the Design Sprint methodology for the company. A new positive dynamic appeared.
- Creation of four typologies of project based on project priorities and on role definition.
- Development of four agile project starts adapted for each project typology. These project beginnings put the adaptation, the team, and the client in the center of the preoccupations and will allow to go in the right direction from the beginning.
- Transition from a linear project development process to an agile project flow.
Summary:
Digitization is considered very important in financial industry as their product basically relies heavily on information and most of the time, physical interaction is not essential part of the service. Bank’s IT costs account for 15–20% of their own expense, and it has been showed that bank improved their profitability when they have more financial innovation. Meanwhile, fintech utilize technology and apply it to financial industry. Since 2016, they started to impose threats to bank, which was revealed by bank’s SEC filings. While competing with fintech is one option bank could pursue, some authors suggest that it is better for them to collaborate with fintech. In response to fintech, more than 70% of financial institutions across the globe have been looking to work together with fintech, either through collaboration, investment, and M&A.
This paper contributes by exploring how fintech approach bank and how the cooperation between fintech and incumbent banks are done, and by doing so, it revealed various ways fintech and bank could work together. Moreover, this paper also contributes by elaborating the differences on their depth of collaboration. This is important as other paper suggests that deeper collaboration could help create more values (including economic value) in inter-organizational collaboration.
Results:
Through this study, initial contact between fintech and bank were revealed to be initiated through third party, other fintech / banking consultant, fintech accelerator involvement, fintech networking event, and fintech’s executive. Fintech firms could help banking institution through collaboration by providing additional product/service offering for bank, assisting bank in product channeling, contributing to bank’s process improvement, and acquiring new customers. Based on these findings, this paper proposed that collaborating with fintech allow bank to tackle digitalization, allow bank to focus on other areas they are really competent, and helping bank tapping into unbanked population market. Depth of collaboration between bank and fintech varies from the system collaboration; operational collaboration; and strategy collaboration which will be elaborated in this study. It is discovered that collaboration depth is linked with how the product/service is branded.
Summary:
This thesis studies the different tools and methodologies that are available to tackle the forecasting issues in the supply chain and transportation industry.
Considering the tiny profit margin of the industry, volumes are key to be profitable.
In that prospect, getting accurate demand forecasts is critical for the management to decide what, when and how many resources to invest in. Non-reliable supply forecast may lead to operational losses and regrettable cash drain.
To achieve the analysis, we have rolled-out three supervised models (Random Forest, Multilayer Perceptron and Lasso Regression) and four statistical models (simple Holt-Winter, double Holt-Winter, SARIMA and Facebook Prophet algorithm).
Then, we have compared the results of each of these models by using the RMSE as the criteria for accuracy.
Results:
The results indicate that the conventional and statistical models outperform the complexity of supervised methods: indeed SARIMA is providing the best accuracy for the daily aggregation.
We note that this output is consistent with the last M4 contest results, that is considered as the most famous forecast competition.
Summary:
Inside Dassault Systèmes’ s CATIA strategy team, I worked on generative design, this new technology reshaping the way we conceive design. Often overused and too abstract, my thesis first aim is to clear up ambiguities and compare this new approach and the traditional CAD approach. This thesis also attempts to draw up the current state of the art of this technology in terms of existing solutions and competitive landscape. Finally, the generative design approach, far from being limited to specific applications, presents exciting characteristics that could benefit many applications. In this direction, I will expose one strategic development proposal to open a new market and extend the generative design scope.
Results:
My work has helped the CATIA strategy team to consolidate its generative design’s strategic referential by providing information about the status and the future perspectives of this technology.
Summary:
This master thesis deals with replenishment policy models in the luxury watch industry. It is based on the need for the watch brands of Richemont International to capitalize the replenishment along the Supply Chain. The different maturity of the brands in terms of current Supply Chain and replenishment policies made this thesis challenging. The purposes of this work are to develop a replenishment policy model from the distribution centers to the point of sales and to identify the gap for each brand between their current replenishment policy model and such a theoretical model. It was led in the context of the Covid-19 crisis so the analysis of the inventory fed strategical talks in the group. The approach was a mix of theory and simulations, with a definition of a theoretical model and the presentation of simulations depending on different parameters.
Results:
The main outcomes are the rationalization of the current replenishment policy of the brands, the development of a replenishment tool and the calculation of the target inventory of the brands. A comparison was first led between the current inventory policy of the brands and the developed model. Then, the target inventory from the brands was compared with the developed inventory to measure the gap and identify which service level is ensured. The main conclusions are the need to for the brands to improve their inventory levels and better match supply and demand. Depending on the brands, it can be achieved either increasing stock to a sufficient level or reducing useless inventory. Finally, the thesis highlighted the gap between current sales and initial forecasts and proposed to reconsider the production plans identifying the gap between current sales and initial forecasts.