This research paper analyzes the role of ethics in data analytics. The acquisition of ever more pervasive—and invasive—details about our shopping and travel habits is one of the unavoidable facts of modern commerce. The most affluent companies devote untold levels of funding to mining the huge amounts of customer preference data to which they are privy in order to target their sales campaigns to prospective buyers with pinpoint accuracy (Kassner). However, the manner in which this data is collected, protected, and disseminated can bear serious, even dire, ethical ramifications. The implications and warnings borne by these considerations are studied and an attempt is made to derive actionable lessons from them.
The term “big data” is often applied to the untold volumes of data that are provided to modern corporations for them to mine in the hopes of targeting ever more effective sales, marketing, and advertising campaigns. The essential flaw with the acquisition and processing of such data is that it is as powerful as it is error-prone.
Modern computers enable huge amounts of data to be subjected to analysis at breakneck speed. While this ostensibly empowers firms to make judicious, even clever, marketing decisions, the power and the laxity of big data analytics also mean that huge errors can be committed with equal facility. It is also the case that people tend to place too much faith in technology. The result of this sad state of affairs is that decisions can be made based upon quickly effectuated summaries and characterizations that essentially operate by striving to reduce the complexity of the human decision-making process to simple-minded processing of message vectors and decision regions in Kotelnikoff hyperspace (Duda & Hart 130ff.).
The Information Accountability Foundation (IAF) sponsored a study of the essential ethical underpinnings of data analytics (“Unified Ethical Frame”). The study addressed seven specific facets of the data analytics process, to wit, collection; processing; analytics; storage and security; governance; usage; and communication. They explored the ramifications of these distinct facets from the perspective of five independent ethical considerations. Specifically, they were concerned that the entire process of collection, processing, and dissemination demonstrate benefit, progressiveness, sustainability, respect, and fairness to all parties whose habits were ostensibly being invaded, monitored, scrutinized, and eventually sold to third parties.
It is appropriate to address each of these five goals from a general perspective before zeroing in on the individual requirements and concerns that they individually engender. Benefit means that big data analytics should demonstrably improve the lives of consumers, not just of the corporate entities that seek to relieve them of their hard-earned monies. Progressiveness means that a continuous culture of improvement should be plainly evident within the data analytics context. Sustainability means that the insights that can be derived from data analytics should apply for the long term: if they fail to do so, it would appear to indicate a serious lapse in judgment. Respect means that the entire process must be both transparent and inclusive rather than appearing palpably to benefit one demographic stratum more than the next. Finally, fairness means that the potential impact of the entire big data life cycle must be carefully evaluated and shown not to favor or disfavor any particular sector, whether economic, societal, ethnic, or what have you.
From a more close-grained perspective, benefit must be clearly identifiable. Not only the scientists who perform in the trenches of data analytics, but also the business executives who rely upon the judgments of these scientists, should be able to ensure both utility and merit from the entire procedure. It seems that data analytics is ethically inappropriate if it cannot be conducted with an expectation of tangible benefit from end to end. In particular, it must clearly deliver value to all parties involved. This refers not only to the organizations who seek to benefit from protracted trend analysis, but also the customers who will eventually be targeted by these organizations as a direct result of their data analytics methodologies, procedures, and mechanisms (Kassner).
Progressiveness implies a culture of perpetual improvement and innovation. To that end, if the improvements in either marketing or variety of goods and services can be achieved in a manner that is less data-intensive and attendantly less invasive, then this goal should obviously be pursued. The organizations that mine and apply big data should always strive to deliver results that are qualitatively superior as well as quantitatively attractant of greater revenues. At the same time, businesses should strive to minimize their usage of such data, obtaining and retaining the least amount necessary that meets their desired objectives. It must be understood that such minimization of data usage not only promotes more sustainability, but also leads to less risk of failed analysis. In particular, the effort to limit the scope and breadth of the data that is acquired for analytics should serve to eliminate covert correlations or insights that could potentially lead to the alienation or disenfranchisement of various market sectors based upon race, nationality, or affine demographic factors (Kassner).
Sustainability of big data over the long term is a vital constituent of any production-level reliance upon its prospective capabilities and promise. Such sustainability actually entails four discrete considerations: data, algorithms, choice of device, and choice of manufacturer. Data sustainability relates to the variety of social data sets to which organizations are privy. It is conceivable that havoc can result when data deriving from both public and provide sources is combined. Moreover, it is critical to consider sourcing insofar as inconsistencies in such considerations as methodology and sample size can affect not only the integrity of the data itself, but also the resultant sustainability of the algorithms applied to it. Speaking of such algorithms, it is paramount that newly identified algorithms apply with a certain degree of longevity, giving organizations faith in their achievement of meaningful rather than sporadic or even random results. Such longevity is influenced by both the collection techniques and the subsequent analytical procedures. Finally, in terms of device- and manufacturer-dependent sustainability, the chief consideration is the lifespan over which the data being collected is applicable. If a company, for example, develops and fields a device that is capable of collection a certain new class of data, and a concomitant rush to use that data as the basis of newfangled algorithms results, it is necessary to harbor a crystalline understanding of what is likely to happen when the product is discontinued, the corporation is sold off, or the data itself is traded to a third party for good and valuable consideration (Kassner).
The collection of data for big data analytics must also be respectful. To the extent that analytics affect individuals as well as organizations that originate, aggregate, or regulate the data, it is crucial to ensure that the data is used with due consideration of the persons who are directly or indirectly affected by it. In particular, the untold volume of real-time data that is being captured from both social media and widely fielded wearable and “internet of things” devices is significantly impacting what has come to be considered the norm for big data analytics. The obvious result of this situation is that even the most seemingly minor decisions can result in decisions or initiatives that engender potentially huge downstream consequences. In most cases, the originators of the data will likely be the ones most impacted by the resultant analytical procedures. The end result is that such data, whether private or semi-private, can end up being unwarrantedly exposed on an altogether public basis (Kassner).
Finally, big data analytics must be eminently fair. It is clear that United States law forbids any manner of discrimination that depends upon such considerations as gender, genetics, age, or racial composition. However, big data analytics are capable of deriving those characteristics from patterns that inhere to data, not even requiring that database fields that specifically or individually reflect those parameters be individually sought or consulted. The fact that discrimination can result from detached data processing rather than from raw data itself is particularly troubling. Such unintended consequences end up plaguing organizations that stand significantly to profit from patterns that they have mined but who find themselves coming up against clear legislative prohibitions. One potential remedial course of action is to implement a fairness test that evaluates how customers would react were all the details on which data analytics decisions are based were fully exposed. If the result would be to strengthen customer relationships, it seems ethical to employ the data. In the alternative case, the use of such data to target sales and marketing decisions strikes one as reprehensible (Kassner).
The combination of these five fundamental principles of ethical data use may lead to a situation that businesses find eminently pragmatic as well as both respectful of customers’ individualities and conservative of such respect. At the same time, the fact remains that big data analytics is such a complex and challenging field that relies upon a multiplicities of business models and emergent technologies. More over, like any other technological business initiative, it relies upon the individuality and respect that inheres to the human intellectual capital that serves as the linchpin and workhorse of the entire data analytics life cycle.
The formal studies that have been commissioned to date purport to address the ethical concerns but pay scant attention to a critical concern. This is the issue of data security. While it may be popular to discuss security in a glib and hand-waving manner, the fact is that data security continues to be poorly understood, witness the ever more frequent hacks and break-ins that periodically compromise millions of sensitive customer records (Bradbury). Unfortunately, the supposed push toward dramatically improved data security has yet to filter down from elite government and private laboratories to the trenches of the big data analytics processing facility. Only time and periodic embarrassments—particularly those that end up costing firms in hard fiduciary terms—can hope eventually to redress these inherent weaknesses in the big data analytics pipeline.Free research paper samples and term paper examples available online are plagiarized. They cannot be used as your own paper, even a part of it. You can order a high-quality custom research paper on your topic from expert writers:
EffectivePapers.com is a professional academic paper writing service committed to writing non-plagiarized custom research papers of top quality. All academic papers are written from scratch by highly qualified research paper writers you can hire online. Just proceed with your order, and we will find the best expert for you!
Bradbury, Danny. “How to Protect Your Business from Hackers.” The Guardian, 23 July 2015. Retrieved from https://www.theguardian.com/media-network/2015/jul/23/protect-business-security-threats.
Duda, Richard O., & Hart, Peter E. Pattern Classification and Scene Analysis. John Wiley & Sons, 1973.
Kassner, Michael. “5 Ethical Principles Big Data Analysts Must Follow.” Tech Republic, 2 January 2017. Retrieved from https://www.techrepublic.com/article/5-ethics-principles-big-data-analysts-must-follow/#googDisableSync.
“Unified Ethical Frame for Big Data Analysis.” Information Accountability Foundation, March 2015. Retrieved from http://informationaccountability.org/wp-content/uploads/IAF-Unified-Ethical-Frame.pdf.