Second, financial controlling functions are significantly based on manual processes, such as the receiving process of invoices and forwarding them to the corresponding business units. Especially in traditional service industries, this is manual paperwork in order to read them but more important to check and proof the validity of these invoices. Adapting natural language processing, this whole process can be automated. By combining existing master data and finance systems, a classification of contents can be done based on cluster analysis for example. Using a semi‐automated solution in a first step, these clusters need to be interpreted in order to assign the related business. Implementing machine learning algorithms, this assignment can be trained and thus in short term be done autonomously. Compared to solutions today, this system would be also tolerant to missing parameters in invoices as the system would know from historical data which parameters have the highest probability to be added in order to fully complete the invoice. This would reduce delays and contractual penalties to a minimum. However, the question will remain if this area won’t be digitalized anyways as supplier will be forced to digitalize their financial outputs, as Amazon proofs. They established this as an assumption for suppliers on their platform.
Autonomous pricing is a third area of potential of autonomous enterprises. Today, prices, for instance of spare parts, are based on expert knowledge and are heterogeneously and decentralized defined in each country. The main problem is that a central market transparency is not present. Analyzing unstructured content, for instance from online platforms, would increase this transparency and will give a central idea about competitive pricing. The challenge is to find 100% matches between web content and internal parts distribution data. Machine learning methods, such as picture recognition, natural language processing or time series analysis [29], allow this matching procedure. Based on this information applications can be designed that define optimal prices in real‐time, multiple times a day. Humans won’t be able to do this independent to their degree of expertise. This scenario is a core business model for online platforms already and will be indispensable to survive in the future competition.
Summarizing, the big challenge is to bridge the gap between developing a new product portfolio and being more cost efficient.
To overcome these challenges, the importance of new technologies, such as machine learning, high speed computing or cloud services has never been higher. The potentials for the automotive industry by investing in these areas are significant: The amount of data will rapidly grow to amounts that overcome most of todays’ architectures,
The capabilities to analyze this amount of data will require machine learning and artificial intelligence techniques to generate the value that exists in this information,
The internal, heterogeneous data sources must be homogenized in order to develop services that go beyond today’s business models,
Internal data must be merged with external data in order to provide services fitting to the concepts of smart cities and the customers’ ecosystem,
Sensor fusion will be a key requirement to establish smart production and predictive maintenance,
Cost efficient business models will require autonomous enterprises.
To remain competitive, current business models need to change according to newly available technologies. Established processes will change towards data driven business models, e. g. the After‐Sales sector in particular. More precisely, current processes will get autonomous. Taking Amazon as an example, autonomous pricing models are needed in order to be competitive in sales, in real‐time. Such platforms are validating and optimizing their portfolio million times each day to increase market share. Automation of business processes have been in the focus and perfected by software companies. In the near future, business processes and functions, such as financially planning and controlling, will be run by machine learning algorithms – autonomously – without human intervention. This will be a completely new level of automation at the highest possible accuracy and quality level. Even decision making, in certain processes will be handled by the machine.
The automotive industry, together with their underlying potential of existing and continuously growing data, has the potential to gather the market leadership. The automation of already existing processes is mandatory to be cost efficient in order to develop the above mentioned capabilities and to follow the disruptive change that requires a new product and service portfolio. Without intelligent and autonomous business models, this will not be possible.
References
1.
Meinzer, S., et al., Translating satisfaction determination from health care to the automotive industry. Service Business, 2016. 10(4): p. 1-35.
2.
Gehrke, L., D. Hajizadeh, C. Kreibich, S. Meinzer, S. Augustine, and J. S. Geffers, Innovation Labs und Aktivitäten der Volkswagen Konzern IT im Zuge der Digitalisierung und Industrie 4.0, in Einführung und Umsetzung von Industrie 4.0: Grundlagen, Vorgehensmodell und Use Cases aus der Praxis, A. Roth, Editor. 2016, Springer: Berlin, Germany. p. 231-246.
3.
Hünecke, P. and M. Gunkel, The influence of After-Sales service determinants on brand loyalty within the premium automotive industry: an empirical comparison of three countries. Service Science, 2012. 4(4): p. 365-381.Crossref
4.
Singularity. Exponential Growth of Computing. 2016; Available from: http://www.singularity.com/charts/page70.html.
5.
Kurzweil, R., How to Create a Mind: The Secret of Human Thought Revealed. 2012, London, England: Viking Adult.
6.
IBM. IBM big data for the automotive industry: Take advantage of new sources of data to reduce costs and increase competitive advantage. 2013 09/10/2016]; Available from: http://www.oesa.org/Doc-Vault/Knowledge-Center/Operational-Performance-Content/IBM-Big-Data-for-Auto-Industry.pdf.
7.
Mearian, L., Self-driving cars could create 1GB of data a second, in Computerworld. 2013, Computerworld.
8.
Rodgers, L., S. Zoepf, and J. Prenninger. Analysing the energy consumption of the BMW ActiveE field trial vehicles with application to distance to empty algorithms. in Transportation Research Procedia. 2014.
9.
Seiberth, G., Wie verändern digitale Plattformen die Automobilwirtschaft?, in Praxis – Industrie 4.0: Wie digitale Plattformen unsere Wirtschaft verändert - und wie die Politik gestalten kann, Ansgar Baums | Martin Schössler | Ben Scott (Hg.): Editor. 2015, Kompendium Digitale Standortpolitik: Berlin, Germany. p. 28-41.
10.
Wedeniwski, S., The Mobility Revolution in the Automotive Industry: How not to miss the digital turnpike. 2015, Berlin, Germany: Springer.Crossref
11.
Tummala, R., et al. New era in automotive electronics, a co-development by Georgia tech and its automotive partners. in 2016 Pan Pacific Microelectronics Symposium (Pan Pacific). 2016. IEEE.
12.
Niculescu, A.I., et al., Designing IDA - An Intelligent Driver Assistant for Smart City Parking in Singapore, in Human-Computer Interaction – INTERACT 2015: 15th IFIP TC 13 International Conference, Bamberg, Germany, September 14-18, 2015, Proceedings, Part IV, J. Abascal, et al., Editors. 2015, Springer International Publishing: Cham. p. 510-513.Crossref
13.
Sinaeepourfard, A., et al. Estimating Smart City sensors data generation: Current and Future Data in the City of Barcelona. in 2016 Mediterranean Ad Hoc Networking Workshop (Med-Hoc-Net). 2016.
14.
Geng, Y. and C.G. Cassandras, A new “Smart Parking” System Infrastructure and Implementation. Procedia - Social and Behavioral Sciences, 2012. 54: p. 1278-1287.Crossref
15.
ADL, Automotive After Sales 2015: Are you ready for the battle?, in Arthur D. Little - Automotive INSIGHT. 2015, Arthur D. Little, Germany. p. 1-6.
16.
Gaiardelli, P., N. Saccani, and L. Songini, Performance measurement of the after-sales s
ervice network - evidence from the automotive industry. Computers in Industry, 2007. 58(7): p. 698-708.Crossref
17.
Philipp Grosse Kleimann, Dorit Posdorf, Alexander Brenner, Swen Beyer, Dr. Ralf Kiene, Torsten Hunstock. Customizing aftersales: Delivering the service that customers really want. 2013 08/28/2016]; Available from: https://de.scribd.com/document/254676885/Roland-Berger-Customizing-Aftersales-20131120.
18.
Singh, S. Future Of Automotive Aftermarket And Car Servicing: Consumers Will Have More Channels To Shop Around. 2015 08/30/2016]; Available from: http://www.forbes.com/sites/sarwantsingh/2015/06/02/future-of-automotive-aftermarket-and-car-servicing-consumers-will-have-more-channels-to-shop-around/#35235c0a4359.
19.
Bandaru, S., et al., Development, analysis and applications of a quantitative methodology for assessing customer satisfaction using evolutionary optimization. Applied Soft Computing, 2015. 30: p. 265-278.Crossref
20.
Chougule, R., V.R. Khare, and K. Pattada, A fuzzy logic based approach for modeling quality and reliability related customer satisfaction in the automotive domain. Expert Systems with Applications, 2013. 40(2): p. 800-810.Crossref
21.
Mavridou, E., et al., Mining affective needs of automotive industry customers for building a mass-customization recommender system. Journal of Intelligent Manufacturing, 2013. 24(2): p. 251-265.Crossref
22.
Mikusz, M., C. Jud, and T. Schäfer, Business Model Patterns for the Connected Car and the Example of Data Orchestrator, in Software Business: 6th International Conference, ICSOB 2015, Braga, Portugal, June 10-12, 2015, Proceedings, M.J. Fernandes, J.R. Machado, and K. Wnuk, Editors. 2015, Springer International Publishing: Cham. p. 167-173.Crossref
23.
Kohl, J., et al., Using multivariate split analysis for an improved maintenance of automotive diagnosis functions, in 15th European Conference on Software Maintenance and Reengineering (CSMR), T. Mens, Y. Kanellopoulos, and A. Winter, Editors. 2011, IEEE Computer Society: Oldenburg, Germany. p. 305–308.
24.
Garcia, E., et al., The evolution of robotics research. IEEE Robotics & Automation Magazine, 2007. 14(1): p. 90-103.Crossref
25.
Scholer, M., M. Vette, and M. Rainer, A lightweight robot system designed for the optimisation of an automotive end-off line process station. Industrial Robot: An International Journal, 2015. 42(4): p. 296-305.Crossref
26.
DWave, Digital pioneering work: Volkswagen uses quantum computers, 2017. [Online]. Available: https://www.dwavesys.com/press-releases/digital-pioneering-work-volkswagen-uses-quantum-computers.
27.
Meinzer, Stefan; Jensen, Ulf; Thamm, Alexander; Hornegger, Joachim; Eskofier, Björn M., Can machine learning techniques predict customer dissatisfaction? A feasibility study for the automotive industry, Artificial Intelligence Research, 2017, Vol. 6 (1), pp. 80-90.
28.
Hofmann, Martin; Neukart, Florian; Bäck, Thomas, Artificial Intelligence and Data Science in the Automotive Industry, 2017. [Online]. Available: https://data-science-blog.com/blog/2017/05/06/artificial-intelligence-and-data-science-in-the-automotive-industry/.
29.
Neukart, Florian: Reverse Engineering the Mind: Consciously Acting Machines and Accelerated Evolution, AutoUni – Schriftenreihe, 1st edition, Wolfsburg, Germany, 2017.
© Springer-Verlag GmbH Germany 2018
Claudia Linnhoff-Popien, Ralf Schneider and Michael Zaddach (eds.)Digital Marketplaces Unleashedhttps://doi.org/10.1007/978-3-662-49275-8_52
52. Successful Data Science Is a Communication Challenge
Martin Werner1 and Sebastian Feld1
(1)Ludwig-Maximilians-Universität München, Munich, Germany
Martin Werner
Email: [email protected]
Sebastian Feld (Corresponding author)
Email: [email protected]
52.1 Introduction
In the last decades, digital marketplaces have seen a deep transformation from providing classical services over a digital channel to providing digital services. Examples for classical services over a digital channel are email service provider like Gmail as an alternative to fax and mail, Internet‐based retailer like Amazon as an alternative to stores, and online trading companies as an alternative to equity traders. Examples for modern digital services are recommendation platforms like TripAdvisor for travel‐related content, communication tools like TeamViewer, online social networks like LinkedIn, or media streaming services like Spotify or Netflix. The most important change in this transformation is represented by the incorporation of added‐value from user data.
This change has led to a situation in which companies are able to collect large amounts of data describing their customers, their behavior, and interaction with the products. Spotify, for example, one of the major music, podcast, and video streaming services, deals with the huge challenge of personalizing their music catalog of over 30 million songs for their more than 60 million active users [1] generating many terabytes of data every day [2]. Netflix, a media production and streaming company, processes more than one billion events per day while petabytes of data are persisted and made available to queries on Amazon’s S3, a web storage service [3]. These examples are located at the core of business intelligence, business analytics, and Big Data. However, a large fraction of this data is unexplored due to corporations not following a sufficient digital strategy, i. e. a company’s sustainable change to digital processes.
An integral part of a digital strategy is a data science strategy – that is the establishment and integration of strategic data science groups in corporations. In this article, we want to discuss what a data scientist is supposed to do and how one can create impact and value from data.
Therefore, we discuss the widely known model defining Data Science as an intersection of mathematics and statistics, computer science, and domain knowledge. We align these three characters with the following four skills to show that a successful data science group must have all of these skills: (1) handling Big Data as a programming challenge, (2) detecting limitations as a statistics and mathematics challenge, (3) awareness and management as a domain expert challenge, and finally (4) usefulness and understanding as an overall challenge.
52.2 The Ten V’s of Big Data
On February 6th, 2001, Doug Laney published a report on data management [4]. In this report, Laney aligns the main challenges for data management into a 3D cube with dimensions Volume, Velocity, and Variety. These three dimensions stand for individual challenges, which are, however, not independent in practice. Usually, a data management problem can be projected into this 3D space giving attributes to the problem helping in identifying challenges, tools, and solutions.
This simple model proved expressive, powerful, and useful for a long time and was often used as one of the “definitions” of the term Big Data. In a recent blog post by Kirk D. Borne [5], this model gets very well commented, also together with an example: the Large Synoptic Survey Telescope (LSST) generates 30 terabytes of imagery every night and has been doing so for the last 10 years. While this clearly is a large volume summing up to 100–200 petabyte over a ten year survey, data velocity is a larger concern: there is one 6 gigabyte image every 20 s. Within 60 s, this image needs to be processed inside the project goals and will generate lots of alerts to be processed further. This is high velocity data. Furthermore, all of this data is processed into a dataset containing 50 billion astronomical objects each with 200,000 attributes. This is high variety. More information on this can be found from the references of the blog article [5].
This list of three V’s has shaped the field of Big Data for a long time and even today, they seem to be the most important aspects of Big Data. However, data scientists are confro
nted with a lot of different problems in the field, some with overlap to the initial 3 V model, some without. In the mentioned blog post, big data influencer Kirk D. Bone presents a list of the 10 V’s of Big Data, which actually covers lot of data science aspects. The viewpoint is widened to include knowledge extraction into the V’s instead of focusing on processing, handling, and organization.
These 10 V’s are given in this section to discuss them later with respect to communication challenges in data science groups. The interesting observation about those 10 V’s is that they are clearly problems, but that different subdomains of Data Science will have different definitions or meanings for the terms. The 10 V’s of Big Data as proposed by Kirk D. Borne are introduced as follows:
Digital Marketplaces Unleashed Page 79