Running Head: W3 Case Studies
W3 Case Studies 2
University of the Cumberland’s
ITS-531-09 Business Intelligence
8th Nov 2019
Table of Contents Introduction 3 Application case 3 1. What do you think about data mining and its implications concerning privacy? What is the threshold between knowledge discovery and privacy infringement? 3 2. Did Target go too far? Did they do anything illegal? What do you think they should have done? What do you think they should do now (quit these types of practices)? 5 Coors Brewers’ Case Study 5 Application case for End of the chapters. 5 1. Why is beer flavor important to Coors’ profitability? 5 2. What is the objective of the neural network used at Coors? 6 3. Why were the Results of Coors’ Neural Network Initially Poor, And What Was Done to Improve the Results? 6 4. What benefits might Coors derive if this project is successful? 6 5. What Modifications Would You Make to Improve the Results of Beer Flavor Prediction? 7 Outline of Modification 8
Target Case Study
Target’s case study aims at analyzing the concept of data collection, mining and its use to predict a customer’s buying behavior and patterns. The firm obtains crucial customer data by assigning each customer a unique guest identification number that is used to track her buying patterns. This case study brings out an important issue; the threshold between data mining and privacy infringement as clearly depicted in the teenager’s case.
1. What do you think about data mining and its implications concerning privacy? What is the threshold between knowledge discovery and privacy infringement?
As highlighted in Target’s case study, privacy is a significant concern in data mining for business purposes. Apart from the data collection technique employed by Target, there exist other data mining applications such as social media and mobile services accessed via the internet that are substantially adopted by individuals in their daily life. The lack of proper privacy protection plan in the process poses a severe threat to individuals. In most instances, the storage and processing of such mined data are usually outsourced to third-party data centers based on the cloud. The privacy concern presents a tremendous obstacle to the full exploitation of the benefits of huge data assets (Analysis and Design of Secured Privacy Data Mining Environment, 2015). Therefore, there is a need to investigate privacy issues in data mining to minimize cases of customer rights infringement.
Privacy is essential to everyone. It allows individuals to decide whether to share any information in question or not. It implicates the supreme sanctity of individual autonomy, and it’s an essential value in any society as it allows people to be individuals. Loss of privacy can be equated with a loss of some traits of humanity. Therefore, the breach of confidentiality by data mining results in a feeling of embarrassment by the offended party. In the case study, the teenage girl is most likely to feel embarrassed because she intended to keep the pregnancy to herself for some time before letting her father know. Generally, privacy invasion makes one vulnerable to all manner of attacks. To maintain the individual’s autonomy and promote a cohesive existence in the society, a boundary between data mining and privacy must be struck.
The threshold between Knowledge Discovery and Privacy Infringement
The data mining for knowledge discovery purposes should only be limited to general information. When sensitive personal information is involved, consent from the individual whose privacy is to be invaded should be sought. Also, assurance of privacy preservation and protection should be guaranteed. Such personal information includes; identification, demographic, financial, and health record data. To obtain such information, the conditions stipulated above should be met and adhered to by the data-mining firm.
The mining of general data such as purchase history (without necessarily noting the client’s name), preference of a particular brand, and general views on the product quality and opinion can be undertaken without many conditions because they don’t affect an individual’s privacy significantly. In our case with Target, the matter of privacy infringement is not coming out clearly because the firm relied on general purchase statistics as a source of their data. What they should have done is to obtain the teenager’s consent on email messaging.
In instances where consent is granted, privacy preservation should be a matter of great interest. The firms can preserve the privacy of sensitive data by using techniques such as; randomization where noise is added to cover the sensitive data records, k-anonymity model, and I-diversity. The two techniques maintain the privacy of the individuals while serving the same purpose of delivering the required results (Analysis and Design of Secured Privacy Data Mining Environment, 2015).
2. Did Target go too far? Did they do anything illegal? What do you think they should have done? What do you think they should do now (quit these types of practices)?
As discussed above, the matter of privacy infringement by Target is not coming out clearly. One may argue that they used general purchase history, but on the other hand, it is questionable whether they obtained the teenager’s consent before sending her emails. In my opinion, they did nothing illegal; they did a good thing by following up on the lady’s pregnancy while advising her on the required purchases. It’s only questionable if the lady had consented to such communication.
They need to conduct their follow-ups differently. They should confirm the customer’s willingness to share sensitive communication with them and through which channels. The follow-ups are helpful to the clients, especially in cases of first-time pregnancies where the ladies don’t know what to purchase and when.
Coors brewers, a British brewing giant, delved into a study to understand beer flavors based on their chemical composition. Such information is vital to Coors in coming up with better flavour’s that suit the customers’ expectations.
The beer industry solely depends on the customer’s tastes and preferences. These (tastes and preferences) are mainly based on the beer flavor. Therefore, coming up with a wide range of flavors assure Coors of an increased market share by giving customers a wide variety to choose from. The consumer preference decisions on beer are guided by the sweetness and dislike for sourness. This illustrates the fundamental role played by flavor on the beer consumption and justifies Coor’s need for creativity to come up with a variety of brands (“The Use of Flavors in Beer and Malt Beverages: A Brief Introduction”, 2019).
The neural network employed was mainly used to obtain the link between inputs and outputs. This was achieved through the exhibition of several blends of required input/output recipes (Sharda, Delen & Turban, n.d.).
3. Why were the Results of Coors’ Neural Network Initially Poor, And What Was Done to Improve the Results?
The first neural network yielded substandard results due to two factors. First, it was unable to extract vital relationships from the data due to the concentration on one product’s quality. This led to a small discrepancy in the data. Second, the functioning of the neural network was influenced by the “noise” generated by inputs that did not affect the flavor at all. This is because only a particular subgroup of the provided inputs influenced the designated beer flavor (Sharda, Delen & Turban, n.d.).
A genetic algorithm was introduced to resolve the challenge. The algorithm could control the various input modifications in reaction to the inaccurate term from the neural network. The algorithm minimized the error term, and later, the switch configuration would recognize the analytical inputs that were probable to forecast the required flavor (Sharda, Delen & Turban, n.d.).
If it becomes effective, Coors will enjoy an increased market share derived from the portfolio diversification. With beer flavors that fit any occasion, venue, and mood, their beers will be the most demanded brands. Ultimately, this translates into an increase in turnover.
Although the analytical method proved successful, modification on the technique by increasing the number of flavor-causal analytics deliberated in the study, in addition to that, flavor active materials and substantial contributors should be taken into account in the complete sensory outline (Sharda, Delen & Turban, n.d.). Besides, the development of sensors that can figure out how to implement diverse fermentation scenarios and using AI solutions to measure the aromas and flavors created by the ingredients will substantially improve the beer quality. Such AI solutions include machine algorithms. The above modifications can be later fine-tuned to come up with a robust software that efficiently predicts and differentiates between different flavors (Laursen & Thorlund, 2017). The schematic diagram below outlines the modification process;
Figure 1: Outline of modification
Analysis and Design of Secured Privacy Data Mining Environment. (2015). International Journal of Science and Research (IJSR), 4(11), pp.149-152.
Laursen, G., & Thorlund, J. (2017). Business analytics for managers. Hoboken, New Jersey: John Wiley & Sons, Inc.
Sharda, R., Delen, D., & Turban, E. Business intelligence, analytics, and data science.
The Use of Flavors in Beer and Malt Beverages: A Brief Introduction. (2019). Technical Quarterly. doi: 10.1094/tq-56-3-0811-01
Vázquez-Araújo, L., Parker, D. and Woods, E. (2013). Comparison of Temporal-Sensory Methods for Beer Flavor Evaluation. Journal of Sensory Studies, 28(5), pp.387-395.
Increase flavor-causal analytes(Analytical method)
Modification of currennt analytical technique
Introduce AI solutions(Machine Algorithms) that measure flavors and aroma
Introducing advanced AI soltions will greatly improve the current progress. Microsoft AI can be utilised.
Fine tune and develop software that differentiates flavors instantly.
This will amplify the current progress with surprising levels of efficiency.