Machine Learning Bridging Between Business And Data Science
Machine Learning: Bridging Between Business and Data Science1
Machine Learning: Bridging Between Business and Data ScienceForeword1. Clarifying the Terms in Data Science1.1 Data Science1.2 Data Mining and Knowledge Discovery In Databases1.3 Machine Learning1.4 Artificial Intelligence2. Machine Learning Workflow By Steps3. Four Groups Of Task That Machine Learning Solves3.1 Classification3.2 Cluster analysis3.3 Regression3.4 Ranking3.5 Generation4. Three Model Training Styles4.1 Supervised learning4.2 Unsupervised learning4.3 Reinforcement learning5. Embarking On Machine Learning5.1 “Business translator” and visionary5.2 Data-driven organizationConclusionFurther Reading2
Machine Learning: Bridging Between Business and Data ScienceForewordIf the past few years hasn’t found you livingHence, this whitepaper is aimed at answeringon a desert island without electricity orpractical questions instead of setting thecommunication with the outside world, you’vevision and evangelizing the trend. This is aboutlikely heard about machine learning (ML). It’san umbrella term data science and how itshard to miss the trend. Every time we talksubfields interact, the main problems thatabout self-driving cars, chatbots, AlphaGo, ormachine learning can solve, and how thesepredictive analytics, we’re discussing someproblems can be translated into the languageimplementation of machine learning techniques.of business. We will also contemplate the mainWhile success stories and evangelists abound,decisions to make concerning talent acquisitionmachine learning hasn’t become the obligatoryand pinpoint the challenges to be considered infor business yet. In the public’s perception,advance. Because we’ve covered data science’salgorithms that are applied in ML are close topotential in articles dedicated to the travel andscience fiction, and rolling out a concrete planindustries, we will only touch on it briefly today.for ML adoption is still a high hurdle.3
Machine Learning: Bridging Between Business and Data Science1. Clarifying the terms in data scienceThe concept of machine learning was firstpattern, machine learning can be effectivelyintroduced back in the 1950s by people whoutilized with thousands of data characteristics.were the remarkable AI pioneers of that time.In 1950, Alan Turing published the “ComputingIn this section, we’ll discuss several fields of dataMachinery and Intelligence” paper thatscience and how they are connected with eachsuggested a famous AI-evaluation test that weother.know today as the Turing Test. In 1959, ArthurLee Samuel coined the term “machine learning.”Many theoretical discoveries that we use weremade at that time. But why are we talking somuch about machine learning and data sciencetoday?Perhaps, the most important difference is the1.1 Data ScienceThe term data science was conceived back in the1960s. While there are many definitions of it, theone which is business-centric was articulated byJohn W. Foreman, the Chief Data Scientist forMailChimp:computational powers and the amount ofdata we can collect and analyze compared toprevious decades. A smartphone that easilyfits in the palm of the hand today can storeand process more data than a mainframecomputer of the ‘60s, which occupied several“Data science is the transformation of data usingmathematics and statistics into valuable insights,decisions, and products”rooms. Instead of relying on thoroughly curatedand small datasets, we can use large andunorganized data with thousands of parametersto train algorithms and draw predictions.The amount and quality of data are whatalso differentiates modern machine learningtechniques from statistics. While statisticsusually rely on a few variables to capture aAs data science evolves and gains new“instruments” over time, the core business goalremains focused on finding useful patterns andyielding valuable insights from data. Today,data science is employed across a broad range4
Machine Learning: Bridging Between Business and Data Scienceof industries and aids in various analyticalhealthcare, analyzing patients’ medical recordsproblems. For example, in marketing, exploringcan show the probability of having diseases, etc.customer age, gender, location, and behaviorallows for making highly targeted campaigns,The data science landscape encompassesevaluating how prone customers are tomultiple interconnected fields that leveragemake a purchase or leave. In banking, findingdifferent techniques and tools.outlying client actions aids in detecting fraud. InPatternRecognitionData Miningand tatisticsData Science Disciplines5
Machine Learning: Bridging Between Business and Data Science1.2 Data Mining and KnowledgeDiscovery in Databasescreating algorithms to extract valuable insightsAs you see from the diagram, all data sciencebasic problem of data mining is to map availablefields are connected with data mining as itdata and convert it into digestible patterns.constitutes the core set of practices within dataData mining is considered to be a part of ascience. The term data mining is a misnomerbroader process called Knowledge Discovery inand doesn’t portray what it stands for. InsteadDatabases (KDD) which was introduced in 1984of mining data itself, the discipline is aboutby Gregory Piatetsky-Shapiro.from large and possibly unstructured data. TheEvaluationData EPAT T E R N STRANSFORMEDD ATAPREPROCESSEDD ATATA R G E T D ATAD ATAKnowledge Discovery in DatabasesWhile it seems that data mining and KDD solely address the main problem of data science, machinelearning adds business efficiency to it.6
Machine Learning: Bridging Between Business and Data Science1.3 Machine LearningThere’s a difference between data miningand machine learning. Machine learning isabout creating algorithms to extract valuableinsights. It’s heavily focused on continuoususe in dynamically changing environmentsand emphasizes adjustments, retraining,and updating algorithms based on previousexperiences. The goal of machine learning isto constantly adapt to new data and discovernew patterns or rules in it. Sometimes it can berealized without human guidance and explicitreprogramming.Machine learning is the most dynamicallydeveloping field of data science today due to anumber of recent theoretical and technologicalbreakthroughs. They led to natural languageprocessing, image recognition, or even thegeneration of new images, music, and texts bymachines. Machine learning remains the maininstrument of building artificial intelligence.1.4 Artificial Intelligencethink and reason as humans do (or approachthis ability). However, with this term so widelyused, we haven’t yet agreed on interpreting the Iin AI. Intelligence is hard to delineate, and waysto define it are numerous. In business language,AI can be interpreted as the ability to solve newproblems. Effectively, solving new problemsis the outcome of perception, generalizing,reasoning, and judging.In the public view, AI is usually conceived as theability of machines to solve problems relatedto many fields of knowledge. This would makethem somewhat similar to humans. However,the concept of artificial general intelligence(AGI) remains in the realm of science fictionand doesn’t yet match existing state-of-theart advancements. Such famous systems asAlphaGo, IBM Watson, or Libratus, which hasrecently beaten humans in Texas Hold’em, arerepresentative of artificial narrow intelligence(ANI). They specialize in one area and canperform tasks based on similar techniques toprocess data. So, scaling from ANI to AGI is thebridge that data science is yet to cross, and thisArtificial intelligence (AI) is perhaps the leastbreakthrough isn’t likely to happen for severalunderstood field of data science. It also standsdecades. While the growing fear that machinesdistinctly apart from the rest. The main ideamay take over many jobs is not unreasonable,behind building AI is to use pattern recognitionthe scenario in which machines dominate theand machine learning to build an agent able toworld is.7
Machine Learning: Bridging Between Business and Data Science1.5 Big dataCollecting large quantities of data doesn’tBig data is also an overly hyped andnecessarily equate with the discovery ofmisunderstood concept. The growth of digitalinsightful patterns in it. The concept of bigtransformation in business allowed for gatheringdata implies discovering patterns in largeincreasingly large datasets that contain various,datasets using the techniques of data miningusually unstructured, records about customers,and machine learning. Why is there so muchemployees, and corporate assets. These relateemphasis on big data today? The popularity ofto demographics, interactions and behaviors,big data among technology evangelists stemsendpoint devices, and literally everythingfrom the recent advancements in computationalthat can be tracked by digital means or inputpower. Instead of using limited subsets of datamanually. However, these unstructured datasetsto discover and extrapolate the results to thearen’t yet big data.entire subject field, we can process all raw data,achieve higher accuracy, and find more hiddendependencies. This requires building high-end“Collecting doesn’t mean discovering.”Sean McClure, Ph.D.Director, Data Scienceat Space-Time Insightinfrastructure capable of computing increasinglylarge sets of unstructured data, then acquiringthe tools and expertise to properly visualize thedata and yield the insights contained in it.8
Machine Learning: Bridging Between Business and Data Science2. Machine learning workflow by stepsSo how do we make algorithms find usefuldescribes how an algorithm processes newpatterns in data? The main differencedata after being trained with a subset of historicbetween machine learning and conventionallydata. The goal of training is to develop aprogrammed algorithms is the ability to processmodel capable of formulating a target valuedata without being explicitly programmed.(attribute), some unknown value of each dataThis means that an engineer isn’t required toobject. While this sounds complicated, it reallyprovide elaborate instructions to a machine onisn’t.how to treat each type of data record. Instead,a machine defines these rules itself relying onFor example, you need to predict whetherinput data. Regardless of a unique machinecustomers of your eCommerce store willlearning application, the general workflowmake a purchase or leave. These predictionsremains the same and iteratively repeats oncebuy or leave are the target attributes that wethe results become dated or need higherare looking for. To train a model in doing thisaccuracy. This section is focused on introducingtype of prediction, you “feed” an algorithmthe basic concepts that constitute machinewith a dataset that stores different records oflearningcustomer behaviors and the results (whetherworkflow.customers left or made a purchase). By learningfrom this historic data, a model will be able toThe core artifact of any machine learningmake predictions on future data.execution is a mathematical model, which9
Machine Learning: Bridging Between Business and Data ScienceMachine Learning WorkflowGenerally, the workflow follows these steps:1. Collect data. Use your digital infrastructure4. Train a model. Use a subset of historic dataand other sources to gather as many usefulto enable the algorithm recognize the patternsrecords as possible and unite them as a dataset.in it.2. Prepare data. Prepare the data to be5. Test and validate a model. Evaluate theprocessed in the best possible way. Dataperformance of a model using testing andpreprocessing and cleaning procedures canvalidation subsets of historic data to understandbe quite sophisticated, but they usually aim athow accurate the prediction is.filling the missing values and correcting other6. Deploy a model. Embed the tested modelflaws in the data, like different representationsinto your decision-making framework as a partof the same values in a column (e.g. Decemberof an analytics solution or let users leverage14, 2016 and 12.14.2016 won’t be treated theits capabilities (e.g. better target your productsame by the algorithm).recommendations).3. Split data. Separate subsets of data to train7. Iterate. Collect new data after using thea model and further evaluate how it performsmodel to incrementally improve it.against new data.10
Machine Learning: Bridging Between Business and Data Science3. Five groups of task that machinelearning solvesIn business terms, machine learning addresses a broad spectrum of tasks, but on higher levels, thetasks that algorithms solve fall into five major groups: classification, cluster analysis, regression,ranking, and generation.3.1 ClassificationClassification algorithms define which categorythe objects from the dataset belong to. Thus,categories are usually referred to as classes. ByCAT Ssolving classification problems, you can addressa variety of questions:Binary classification problems Will this lead convert or not?DOGS Is this email spam or not? Is this transaction fraudulent or not?And, multiclass problemsBinary classification Is this apartment in New York, San Francisco, orBoston? What is pictured: a cat, a dog, or a bird? Which type of product is this customer morelikely to buy: a laptop, a desktop, or asmartphone?11
Machine Learning: Bridging Between Business and Data ScienceAnother highly specific type of classification taskis anomaly detection. It’s usually recognizedas the one-class classification because theNORMALDI S T R I BU T I O Ngoal of anomaly detection is to find outliers,O UT L I ERSunusual objects in data that don’t appear in itsnormal distribution. It can solve these types ofproblems: Are there any untypical customers in our dataset? Can we spot unusual behaviors among our bankclients? Does this patient deviate from the rest accordingAnomaly detectionto the records?3.2 Cluster analysisThe main difference between regularclassification and clustering is that the algorithmis challenged to group items in clusters withoutpredefined classes. In other words, it shoulddecide the principles of the division itselfwithout human guidance. Cluster analysisis usually realized within the unsupervisedlearning style, which we will talk about in aminute. Clustering can solve the followingproblems: What are the main segments of customers weCluster analysis (estimated number of clusters: 3)have considering their demographics andbehaviors? Is there any relationship between default risks ofsome bank clients and their behaviors? How can we classify the keywords that people useto reach our website?12
Machine Learning: Bridging Between Business and Data Science3.3 RegressionRegression algorithms define numeric targetvalues instead of classes. By estimatingnumeric variables, these algorithms are usedin predicting product demand, sales figures,marketing returns, etc. For example: How many items of this product will we be ableto sell next month? What’s will the airfare be for this destination? What’s going to be the rental price for this house?Linear regression3.4 RankingRanking algorithms decide the relativeimportance of objects (or items) as related toother objects. The most well-known exampleto rank pages on the search engine resultspage. Ranking algorithms are also applied byFacebook to define which posts in a news feedare more engaging to users than others. Whatother problems can ranking address? Which movies this user will enjoy the most? What hotels will be on the most-recommendedlist for this customer?Predicted Ratingis PageRank, which is heavily used by Google12435PopularityMovie recommendation ranking How should we rank products on a search pageof an eCommerce store?13
Machine Learning: Bridging Between Business and Data Science3.5 GenerationGeneration algorithms are applied to generatetext, images, or music. Today they are used insuch applications as Prisma that converts photosto artwork-style images, or WaveNet by DeepMindthat can mimic human speech or create musicalcompositions. Generative tasks are morecommon for mass consumer applications, ratherthan predictive analytics solutions. That’s whythis type of machine learning has big potentialfor entertainment software. What type of tasksare in the realm of generative algorithms?Image converted to artwork using “The Great Waveoff Kanagawa” piece of art Turn photos into specific style of painting. Create text-to-speech applications for mobilevoice assistants (e.g. the Google assistant). Create music samples of one style or that arereminiscent of a particular musician.To meet these tasks, different model training approaches (or training styles) are used. Training is aprocedure to develop a specific mathematical model that is tailored to dependencies among valuesin historic data. A trained model will be able to recognize these dependencies in future data andpredict the values that you look for. So, there are three styles of model training.14
Machine Learning: Bridging Between Business and Data Science4. Three Model Training StylesChoosing training styles depends on whetherpopular approach utilized in business. Foryou know the target values that should beexample, if you choose binary classification tofound. In other words, you can have trainingpredict the likelihood of lead conversion, youdatasets where the target values are alreadyknow which leads converted and which didn’t.mapped and you just want the algorithmYou can label the target values (converted/notto predict these exact values in future data.converted or 0/1) and further train a model.Or your goal may be to figure out hiddenSupervised learning algorithms are also usedconnections among values. In the latter case,in recognizing objects on pictures, in definingtarget values are unknown both for historic datathe mood of social media posts, and predictingand future data. This difference in goals impactsnumeric values as temperature, prices, etc.the training style choice and defines whichalgorithms you choose.4.1 Supervised learning4.2 Unsupervised learningUnsupervised learning is aimed at organizingdata without labeled target values. The goalSupervised learning algorithms operate withof machine learning, in this case, is to definehistoric data that already has target values.patterns in values and structure the objectsMapping these target values in training datasetsby similarities or differences. In classificationis called labeling. In other words, humans telltasks area, unsupervised learning is usuallythe algorithm what values to look for and whichapplied with clustering algorithms anddecisions are right or wrong. By looking at aanomaly detection. These models are useful inlabel as an example of a successful prediction,finding hidden relations among items, solvingthe algorithm learns to find these target valuessegmentation problems, etc.in future data. Today, supervised machineFor example, a bank can use unsupervisedlearning is actively used both with classificationlearning to split clients into multiple groups.and regression problems as target values areThis will help to develop specific instructionsusually available in training datasets.for dealing with each group. UnsupervisedThis makes supervised learning the mostlearning techniques are also employed in15
Machine Learning: Bridging Between Business and Data Scienceranking algorithms to provide individualizedis employed by the Tesla autopilot along withrecommendations and in generative tasks.supervised learning techniques. The style isutilized when the autopilot is on and a driver4.3 Reinforcement learningReinforcement learning is perhaps the mostsophisticated style of machine learning andis inspired by game theory and behavioristpsychology. An agent (an algorithm) must makedecisions based on input data and then be“awarded” or “punished,” depending on howsuccessful these decisions were. By iterativelyfacing awards and punishments, the agent altersits decisions and gradually learns to achievebetter results.Reinforcement learning techniques today areactively used in robotics and AI development.A well-known AlphaGo algorithm by DeepMindused reinforcement learning to estimate themost productive moves in the ancient game ofcorrects its decisions.However, in business computing, reinforcementlearning is still hard to apply as mostalgorithms can successfully learn only withinthe unchanging framework of rules, goals, andworld circumstances. That’s why today’s manymodern reinforcement learning advancementsare tethered to games like Go or old Atarititles where these three parameters arestable. Another problem of reinforcementlearning is the longevity of learning cycles. Ingames, the time between the first decisionand achieved points is relatively short, while inreal-life circumstances the time to estimate howsuccessful the decision was may take weeks.Go instead of enumerating all possible boardcombinations. Allegedly, reinforcement learning16
Machine Learning: Bridging Between Business and Data Science5. Embarking on machine learningPredictive analytics and machine learning are still terra incognita for most businesses. Although theevolution of machine learning tools seems impressive, capturing the business value is still challenging.Companies stumble over talent acquisition barriers, internal leadership difficulties, and, last but notleast, the rigidity of overregulated corporate culture. It’s relatively easy to theorize about the greatpotential of big data–which looms large in the media–but the reality is that the number of companiesplanning to invest in big data sank from 31 to 25 percent in 2016. On the other hand, the investment inbig data is generally up thanks to big players. This means that the competitive gap only increased forsmaller or less flexible businesses.In this section, we’ll talk about the most critical decisions that should be made on executive level toovercome these barriers and align with competition.5.1 “Business translator” andvisionarymanage efficient data processes. Although youProper analytics and data science leadership ismissing links without an analytics leader, thisthe greatest barrier to achieving data-drivenculture. According to the McKinsey GlobalInstitute survey, 45 percent of companies arestruggling to set the right vision and strategy fordata and machine learning. Consider this–thechallenges of talent acquisition are well-known:Data science talent is scarce and expensive bothin terms of compensation and retention. Whilefinding a data scientist is hard, finding ananalytics leader is even more difficult, accordingto the survey. Ironically, this role is critical tocan introduce some machine learningimplementations and compensate for a fewapproach is destined to remain responsiverather than proactive.The skillset of this “business translator,” or chiefanalytics officer (CAO), is a multidisciplinarybridge between business values and datascience capabilities. The person should take thelead and reconcile the efforts of the informationtechnology department, data science, marketing,finance, and stakeholders to build and develop adata strategy.17
Machine Learning: Bridging Between Business and Data ScienceChief Analytics Officer Engagement FieldAnother important mission of an analyticsdecades. And many–like reinforcement learning–leader is a visionary one. It implies foreseeinghave yet to find their implementations beyondbusiness application potential in new dataprominent labs like DeepMind. By capturingscience research works before they arethese advancements early and finding wayswidely adopted. Most of the machine learningto convert them into business use a businesstechniques that have met business demandstranslator can keep the organization ahead oflately have been known in data science forthe competition.18
Machine Learning: Bridging Between Business and Data ScienceHowever, analytics specialist acquisition won’tintuition and experience, which made thembe simple. The current mismatch between theprofessionals prior to predictive analyticsdemand for senior analytics positions and talentbursting on the scene. The role of an analyticssupply stands at 5:1. And if recruitment fails, thisleader (or CAO) and other C-level executives isdivision implies finding and training an analyticsto educate employees and foster the innovation.expert internally. The best-fit opportunity, inThis is the reason why communication andthis case, is to engage a person who has bothpresentation skills are preferred qualities for atechnical and domain business background.data scientist.Sometimes, this role can be obtained by achief technology officer, a data scientist whoSiloed data. The siloed structure oftransitions into management, or even a chiefdepartments is another barrier to building aexecutive officer. That decision depends on thedata-driven organization. Access to data canorganization size.be either overregulated or warily guarded bydepartments that may want to keep the data5.2 Data-driven organizationA data scientist alone can only be effectivewithin a fertile corporate environment.Introducing a machine learning initiativeshould be supported and understood on allorganizational levels. With each new technologycoming, not only training is required, but alsoimmense effort in evangelizing change. If youplan to use machine learning as a support todecision-making or as a lever to make importantdecisions, most likely this way of thinking isthey collect to themselves. By combating thisbehavior you can achieve much better results inacquiring more useful data.Anonymized data. Sometimes regulations areimposed legally in such businesses as bankingor insurance and data can’t be easily shared. Inthis case, all values in data can be turned intoanonymized numbers at the data preparationstage. Thus sensitive business or customerdetails won’t be revealed.going to face reasonable resistance. Peopleare used to making decisions based on their19
Machine Learning: Bridging Between Business and Data ScienceConclusionThis paper isn’t intended to be exhaustive and shouldn’t be considered as a playbook for your emergingmachine learning initiative. While there is much to explore, we rather suggest using this white paper asa guide to evaluate your strategy.The bottom line problem for business today is to understand how and when this strategy is going to berealized to keep up with the pace of change that machine learning and predictive analytics can provide.The modern era of business decisions will put ahead of the competition those who can make the bestuse of data they collect.Further Reading1. Data Smart: Using Data Science to Transform Information into Insight, Per John W. Foreman - https://books.google.cat/books?id CfjpAQAAQBAJ&pg PR14#v onepage&q&f false2. From Data Mining to Knowledge Discovery in Databases, Fayyad, Piatetsky-Shapiro & Smyth, 1996 - le/viewFile/1230/11313. A Collection of Definitions of Intelligence, Shane Legg, Marcus Hutter, 2007 - https://arxiv.org/pdf/0706.3639.pdf4. o-very-different-beasts-sean-mcclure-ph-d-?trk mp-reader-card5. http://www.gartner.com/newsroom/id/34661176. The Age of Analytics: Competing in a Data-Driven World, 2016 - ting-in-a-data-driven-world7. 85f6edf5f718. ls-data-scientist.html9. lary-survey20
Machine Learning: Bridging Between Business and Data ScienceAbout AltexSoftAltexSoft is a Technology & Solution Consulting company co-building technology products to helpcompanies accelerate growth. The AltexSoft team achieves this by leveraging their technical, processand domain expertise and access to the best price-for-value Eastern European engineers. Over 100US-based and 200 worldwide businesses have chosen the company as their Technology ConsultingPartner.US Sales HQGlobal HQ701 Palomar Airport Road,32 Pushkinskaya Str.,Suit 300, Carlsbad, CA 92011Kharkiv, Ukraine 61057 1 (877) 777-9097 38 (057) 714-153721
2. Machine Learning Workflow By Steps 3. Four Groups Of Task That Machine Learning Solves 3.1 Classification 3.2 Cluster analysis 3.3 Regression 3.4 Ranking 3.5 Generation 4. Three Model Training Styles 4.1 Supervised learning 4.2 Unsupervised learning 4.3 Reinforcement learning 5. Embarking On Machine Learning 5.1 "Business translator" and .
Each level of Girl Scouting has its own unique bridging award patch. Bridging Ceremonies Bridging ceremonies often utilize a bridge as girls take literal steps toward the future. For Girl Scouts, the act of crossing the bridge is both a physical and symbolic step. Bridging ceremonies can: Include troops, groups, or individuals
(bridging atoms) The orientation can be random, leading to an amorphous structure. Some oxygen atoms will be bonded to only one silicon atom (non-bridging atoms). The relative amounts of bridging to non- bridging determines the "quality" of the oxide. If all oxygen atoms are bridging, then a regular crystal structure results - quartz. SiO
Network Bridging Setup Guide www.cetoncorp.com Network Bridging Setup Guide 2013 Network Bridging (a.k.a. Network Tuners) is compatible with PCs running Windows 7 with Media . Selecting the "configure manually" option will allow you to choose any available wired or wireless NIC. www.cetoncorp.com Network Bridging Setup Guide 2013 4 .
decoration machine mortar machine paster machine plater machine wall machinery putzmeister plastering machine mortar spraying machine india ez renda automatic rendering machine price wall painting machine price machine manufacturers in china mail concrete mixer machines cement mixture machine wall finishing machine .
between plies of dissimilar orientation, so fiber-bridging does not occur. Therefore, in order to be useful in structural modeling, expressions relating the delamination growth rate and strain energy release rate must account for the effect of fiber-bridging. Fiber-bridging under quasi-static loading can be quantified as a delamination .
Machine learning has many different faces. We are interested in these aspects of machine learning which are related to representation theory. However, machine learning has been combined with other areas of mathematics. Statistical machine learning. Topological machine learning. Computer science. Wojciech Czaja Mathematical Methods in Machine .
Bridging the Gap Across the Sister Islands FEATURES OF THIS EDITION H O U S E O F A S E M B L Y V I R G IN S L A N D S 1st Quarter 2022 BRIDGING THE GAP Message: Bridging the Gap Across the Sister Islands (Hon. Shereen Flax-Charles) Special Feature: BVI NPO Collaboration Service Day on ANEGADA Sister Islander of the Quarter: Guess Who?
A/ B. COM - SEMESTER I – GENERAL ENGLISH (2019- 20) University Paper Style (total 4 questions, 70 marks, 2.30 hours) Unit/s Topic/s No MarksQuestion style I Lessons Beautiful Minds (Gujarati Medium) Pinnacle (English Medium) Q. 1. 1 to 3 (a) Answer in brief - 3/5 (b) Write a short note - 1/3 (09) (08) II Q. 2. Poems 1 to 3 (a) Answer in brief - 3/5 (b) Write a short note - 1/3 (09) (08) III .
