Electric Pickaxes: Data Minings Benefits, Concerns, and Limitations
Jason Faulk June 5, 2014 Keagan Samuel CIS 500: Information Systems Decision-making
ELECTRIC PICKAXES 2 Electric Pickaxes: Data Minings Benefits, Concerns, and Limitations Humanity has always wanted to know what the future holds and what others were thinking. Crystal balls, tarot cards, and tea leaves have all been employed with mixed results. If someone could figure out how to predict the future- the next big trend, what way the stock market will head- then that person could make a lot of money, and if a company could just figure out how to read their customers minds then their stock would skyrocket. Since mind reading and fortunetelling are still impossible, companies have found a new way to predict the future and see what their customers are thinking- data mining. Data mining offers unique benefits to businesses through predictive analytics, associations discovery, web mining and clustering, but depends on complex algorithms that may have some reliability flaws, and raises privacy concerns among the customers whose data is being mined.
Tools in the Mine Cart: The Techniques Used in Data Mining There are several different techniques that can be used in data mining to provide unique benefits to businesses. These techniques allow companies to predict trends, estimate performance, and offer the right products in a way that companies that dont use data mining cannot offer. By utilizing these techniques, companies are able to gain a distinct advantage over their competitors. Almost every field can benefit from data mining techniques and using them has become almost required in certain fields. The first technique is using predictive analytics to find meaning in customer actions. Predict analytics was developed as a way for companies to use data to make educated guesses about the future. In their book, Information Technology for Management, Turban and Volonino describe predictive analytics as the branch of data mining that focuses on forecasting trends ELECTRIC PICKAXES 3 (e.g., regression analysis) and estimating probabilities of future events (2011). They also list the top five reasons companies use predictive analytics, which are: improving customer loyality and satisfaction, improving growth strategies, discovering process inefficiencies, identifying and responding to adverse situations, and optimizing resources (2011). Secondly, data mining uses association discovery to find what else customers might want to buy based on what items are typically purchased together. Association discovery uses data rules- if/then rules that predict what seemingly unrelated item a customer might buy based data mined from past purchase history. Kaur, Bedi, and Gupta in their article, Review of Association Rule Mining, explain that association rules mining is an important data mining task that used to find out correlations, association between a set of trans-actions in the databases (2014). By using this technique, retailers can determine which two random items are usually bought together. This knowledge allows them to place these items close to each other encouraging sales or offer coupons on one for the other item. Another techniques used in data mining is web mining to see what can be gleaned from the web activities of customers. Turban and Volonino define it as the application of data mining techniques to discover actionable and meaningful patterns, profiles, and trends from Web resources (2011). Everyone is familiar with ads on Google or Facebook that seem targeted specifically at you. These ads are created by those companies mining your web activity and web purchases to determine what ads you would be more likely to click. Companies can also mine their own traffic to see how successful ad placement and web design are on their site. Finally, data mining uses clustering to bring like customers and determine what that cluster likes. Clustering is simply bringing together groups of like things that are different from other groups of like things. Companies may sort their customers by age, race, gender, education, ELECTRIC PICKAXES 4 income or a variety of other things. They can then analyze the purchases or web habits of these groups to determine their success among a certain demographic or which cluster they should target. It also helps companies know what kind of customers they have and how they should market to those customers.
Keystone or Weak Link: The Importance of Algorithms in Data Mining Data mining depends on complex algorithms that while tested and trustworthy cannot predict every action an autonomous human being may make or count for variables that are unpredictable. An algorithm is defined by Turban and Volonino as a set of rules instructions for solving a problem in a finite number of steps (2011). Algorithms can be simple or extremely complex. The more complex an algorithm the more factors it takes into account and the more specific the answer produced. This allows companies to look at a large group of data, fator in many variables and produce a specific answer. Of course the two major problems with algorithms are changing variables and unaccounted for variables. When a variable changes suddenly, an algorithm might not catch it in time and produce obsolete data. When a variable that was not accounted for suddenly has a major impact on the data being mined, then the algorithm may even be rendered useless. Finally algorithms can only make reasonable and logical predictions and sometimes people are anything but reasonable and logical.
Questionable Digging: The Privacy Concerns of Customers While data mining provides a wealth of data to companies, some consumers are worried about their privacy when every click, swipe, tweet, and status update is mined for information. It has already been mentioned that companies like Google and Facebook mine the data they gather ELECTRIC PICKAXES 5 from peoples web activities, but these days they also have access to users private emails and instant messages. People are beginning to question the use of their private communications to produce web ads and privacy for electronic communication is being pushed. Some consumers are even fighting back legally. A recent lawsuit was filed against Google by California students, claiming that Google mined their private education files to produce web ads targeted at them (Neal, 2014). Consumers are not only getting wise online but also taking notice of how things like point and membership cards are really just tools for companies to collect data on their customers. As more people become aware of the scope of data mining, many are choosing to opt out or avoid ways their data can be tracked. Consumers are also beginning to backlash against companies that mine data to aggressively.
In Conclusion In conclusion, data mining is a useful tool that can offer businesses distinct advantages over their completion. Data mining uses a variety of techniques to predict if future trends, likely purchases, characteristics of customers, and successfulness of websites. While the algorithms used in data mining are usually dependable, rapid changes in variables and unpredictable variables can cause obsolete or untrustworthy data. The biggest threat to data mining is its privacy concerns. Customers are beginning to feel uncomfortable with being studied constantly. ELECTRIC PICKAXES 6 References Kaur, Bedi, and Gupta. (2014) Review of association rule mining. Available at http://www.ijater.com/Files/dba82d0e-c39c-4ab6-9629- b9951bffe9eb_IJATER_15_03.pdf Neal, R.C. (2014) Google sued for data-mining: California students claim violation of educational privacy. Available at http://www.ibtimes.com/google-sued-data-mining- california-students-claim-violation-educational-privacy-1562198 Turban, E., & Volonino, L. (2011) Information technology for management (8th ed.). Hoboken, NJ: John Wiley & Sons.