OverviewThe late 1900s heralded in an era of tools formaking better-informed decisions in business.Commercial products became available underthe umbrella terms of decision support systems(DSS), group decision support systems (GDSS),executive information systems (EIS) and expertHOME STUDENT RESOURCES HELP STUDENT HELPANUSHA CLOASGUOLUAT 10systems (ES). Most commercial ES were rulebasedsystems, whereby both the knowledgeand the problem-solving procedures werestored in the form of rules. ES providedexplanatory capabilities, which were essential tothe transfer of expertise and problem solving(Curry, 1993). However, dramatic advances inthe !elds of machine learning, statisticalmodelling and data mining (propelled byadvances in storage technology) have all butreplaced last century’s tools.Today, experts in data science, arti!cialintelligence (AI) and big data are among themost sought-after roles in organisations from allindustry sectors as well as government(Davenport & Patil, 2012). In line with Laureate’smission of giving our students an edge inemployability, this module discusses practicalapplications of data-driven decision tools (Das,2014) and integrates concepts from previousmodules to help understand where these tools!t into decision-making.Data science, AI and big data have di”erentroots in their own distinct data-driven traditions.Arrttii!cciiall IInttelllliigencce is enabling computers tothink. For example, the aforementioned (andsomewhat obsolete) rules-based Expert Systemsform part of AI. Macchiine LLearrniing is a subsetof AI, which uses statistical tools to learn fromdata. In 2006, Deep LLearrniing has emerged asa part of machine learning, which employsmulti-layer neural networks. This is the areawhere most advances have been made in AI. Forexample, when Google’s AI is commonlyexample, when Google’s AI is commonlyreferred to in the literature, the reference ismostly to Google’s deep learning capabilities.Datta Scciiencce is a di”erent discipline from AIalthough it has some overlap with AI andmachine learning (less so with deep learning). Itfollows an interdisciplinary approach, which liesat the intersection of maths, statistics, softwareengineering and systems thinking. Data Sciencedeals with data collection, data cleaning,analysis, visualisation, model creation, modelvalidation, prediction, designing experiments,hypothesis testing and much more. The aim ofall these steps is to derive insights from data.However, the amount of data collected bycompanies today is so vast it creates a large setof challenges regarding data acquisition,storage, analysis and visualisation. Biig Dattainvolves four ‘V’s – Volume, Variety, Veracity andVelocity – which requires it to be treateddi”erently from conventional data as follows:Volume: The amount of data involvedhere is so huge, that it requiresspecialised infrastructure to acquire,store and analyse it. Distributed andparallel computing methods areemployed to handle this volume of data.Variety: Data comes in various formats;structured or unstructured. Structuredmeans neatly arranged rows and columnsin so-called relational databases.Unstructured means that it comes in theform of raw text, videos and images,which requires di”erent databasesystems than relational ones.systems than relational ones.Veracity: Data needs to be correct for it tobe meaningful. The industry term for thisis “garbage in [bad data], garbage out[wrong conclusions]”. Care needs to betaken to make sure the data captured isaccurate and data integrity is maintained— especially, as the data increases involume and variety.Velocity: According to IBM (2016), 90% ofdata in the world was created in theprevious two years alone (Loechner,2016). This velocity of informationgenerated is bringing its own set ofchallenges. For some businesses, realtimeanalysis is crucial. Any delay willreduce the value of the data and itsanalysis for business.The jargon of commercial decision-making toolsin the marketplace can be positioned ondescriptive, diagnostic, predictive or prescriptivedimensions as depicted in Figure 1 below.FFiig 1:: Datta–drriivven ttoollss fforr decciissiionmakkiing(Kaduk, 2016)These product concepts all aspire to convergetowards the top of the DIKW pyramid (referModule 1) in a multidisciplinary approach asshown in Figure 2 below:FFiig 2:: Convverrgencce off ccommerrcciiall ttoollssttowarrdss tthe ttop off tthe DIIKW pyyrramiid(Press, 2016)Such technologies are clearly reinvigorating arational decision-making model, which mayleave us vulnerable to algorithmic & data biasesas well as overcon!dence vis-à-vis blind spots(refer Module 2). Consider this: human choicesand biases a”ect errors already in thetechnology factory. This causes Type I (falsepositives) and Type II errors (false negatives oropportunity cost), which we discuss in thismodule. Therefore, we reinforce the importanceof intuition and balanced decision-making withdata-driven tools. Blind trust in (or outrightrejection of) data-based technology is aconsequence of several myths (Kelleher &Tierney, 2018):We can let data tools run over dataWe can let data tools run over datawithout skilled human oversight atseveral stages of the process;Data science = big data = arti!cialintelligence;Data tools are easy to use & do notrequire deep domain expertise;Data tools pay for themselves quicklyeven in the absence of a well-understoodbusiness problem and appropriate data.The last point above about the creation of databasedtechnology takes us back to groupdecisionsfrom Module 3. Consider “group think”or a lack of diversity in the data science factory,for example. This causes a statistical error,which is then compounded by a decisionmaker’sown judgement, interpretation andintuition bias when decisions are made from thetool’s !ndings.When data-driven technologies are !nally rolledout into the workplace, several changemanagement challenges arise as depicted inFigure 3 below (also refer Soft SystemsMethodology in MGT603: Systems Thinking):FFiig 3:: Change managementt & grroupss iin AIIttrranssfforrmattiion(Fountaine, McCarthy, & Saleh, 2019)For ethical issues involved with data-driven toolsrecall the learnings from Module 4. Supervisedlearning-based data tools are perpetuating“what has been”, meaning the data is leaningtowards keeping the status quo. But what if thestatus quo is not the desirable state? With AI welearn from past data and then deduce the futurefrom the past – and thus AI keeps things the waythey have been. With unsupervised learningbaseddata tools, on the other hand, these toolskeep learning (and make their own decisionsfrom new data) after they have left the data toolfactory. Data Science resolves this throughhuman interaction in problem solving in aconsultative process, but AI does not go as far tothe same degree (Chowdhury & Mulani, 2018):Data is not objective, is it re#ective of preexistingsocial and cultural biases?AI can be a method of perpetuating bias,leading to unintended negativeconsequences and inequitable outcomes.The !eld of AI ethics draws aninterdisciplinary group of lawyers,philosophers, social scientists,programmers, and others. Figure 4 belowdepicts why this is required, consideringthe simple data-based decisions made byAI.FFiig 4:: Whatt AII decciidess wiitthoutt rregarrd ttoetthiiccss(Ng, 2016)ReferencesChowdhury, R., & Mulani, N. (2018, October 24).Auditing algorithms for bias. HarvardBusiness Review. Retrieved fromhttps://hbr.org/2018/10/auditingalgorithms-for-biasCurry, D. J. (1993). The new marketing researchsystems: How to use strategic databaseinformation for better marketingdecisions. Somerset, NJ: John Wiley &SonsDas, S. (2014). Computational business analytics.Boca Raton, FL: CRC PressDavenport, T., & Patil, D. J. (2012, October). DataScientist: The sexiest job of the 21stcentury. Harvard Business Review, 90(10), 70–76Fountaine, T., McCarthy, B., & Saleh, T. (2019,July/ August). Building the AI-poweredorganization. Harvard Business Review,62–73Kaduk, T. (2016, December 14). 4 stages of DataAnalytics maturity: ChallengingGartner’s model. Retrieved fromLinkedin:https://www.linkedin.com/pulse/4-stages-data-analytics-maturitychallenging-gartners-taras-kaduk/?trackingId=OB0BeGS2rIMMauGR56UIzA%3D%3DKelleher, J. D., & Tierney, B. (2018). Data Science.Cambridge, MA: The MIT PressLoechner, J. (2016, December 22). 90% of today’sdata created in two years. ResearchBrief. Centre for media Research.Retrieved fromhttps://www.mediapost.com/publications/article/291358/90-of-todays-data-created-in-twoyears.htmlNg, A. (2016, November 9). What arti!cialintelligence can and can’t do right now.Harvard Business Review. Retrievedfrom https://hbr.org/2016/11/whatarti!cial-intelligence-can-and-cant-doright-now
What Students Are Saying About Us
.......... Customer ID: 12*** | Rating: ⭐⭐⭐⭐⭐"Honestly, I was afraid to send my paper to you, but splendidwritings.com proved they are a trustworthy service. My essay was done in less than a day, and I received a brilliant piece. I didn’t even believe it was my essay at first 🙂 Great job, thank you!"
.......... Customer ID: 14***| Rating: ⭐⭐⭐⭐⭐
"The company has some nice prices and good content. I ordered a term paper here and got a very good one. I'll keep ordering from this website."
