BIG DATA on a car’s location, acceleration, braking,
and distance traveled. Owners who install
3Vs model by Laney the device can get a 10 to 15 percent discount on their policy. Privacy activists, ● Volume however, fear the technology is ripe for ● Variety abuse.” ● Velocity of data DIMENSIONS OF BIG DATA “datasets, which could not be captured, managed, and processed by general ➔ Volume: Magnitude of the data that computers within an acceptable scope”. is being generated and collected. ➔ Velocity: Rate of the generation of McKinsey Global Institute – “datasets data. Big data should be processed whose size is beyond the ability of typical and analyzed in real-or-near-time to database software tools to capture, store, make informed decisions. manage, and analyze”. ➔ Variety: Refers to different types of data being generated and captured. International Data Corporation (IDC ) “big They extend beyond structured data data technologies as new generation of and fall under the categories of technologies and architectures, designed to semi-structured and unstructured economically extract value from very large data. volumes of a wide variety of data, by enabling high-velocity capture, discovery, ADDED Dimensions of Big DATA and/or analysis”. ➔ Veracity: unreliability associated with the data sources. Huge size of unstructured data produced by ◆ SENTIMENT ANALYSIS high-performance heterogenous group of using social media applications that spans from social network to scientific computing applications. ➔ Variability: Inconsistency in the big data velocity leads to variation in The dataset range from a hundred flow rate of data. gigabytes to zetabytes that is beyond the capacity of existing data management tools ➔ Low-Value density: Data in its to capture, store, manage and analyze. original form is unusable. Data is analyzed to discover very high Examples of Big Data value. Ex. logs from website cannot be used in its initial form to obtain The following are summary examples of business value. It must be analyzed how Big Data is being utilized: “Instead of to predict the customer behavior. ‘find my iPhone,’ some auto insurance companies are offering a service that may enable parents to ‘find my teenager.’ Progressive Insurance, for example, offers the Snapshot, a tracking device that reports How to Measure Big Data and What’s a 4. Analyze - From the completed BYTE? surveys, the data analysts ★ A byte = one character, or a grain of discovered that an employee’s sand experience with certain processes ★ A kilobyte = a sentence, or a couple was a key indicator of overall job of pinches of sand • satisfaction. ★ A megabyte = a 20-slide PowerPoint 5. Share - This process gave show, a small book, or a tablespoon managers an opportunity to of sand communicate the results with the ★ A gigabyte = 10 yards of books on a right context. As a result, they could shelf, or a shoebox full of sand • have productive team conversations ★ A terabyte = 300 hours of good- about the next steps to improve quality video, a tenth of the Library employee engagement. of Congress, or a playground 6. Act - The last stage of the process sandbox • for the team of analysts was to work ★ A petabyte = 350,000 digital with leaders within their company pictures, or a mile-long stretch of and decide how best to implement beach • changes and take actions based on ★ An exabyte = half the information the findings. generated worldwide in 1999, or a beach stretching from Maine to North Carolina • DECISION MODELS ★ A zettabyte = unimaginable, or a beach as big as all the coastlines in ➔ An abstraction or representation of a the world real system, idea, or object ➔ Captures the most important features SIX STEPS ➔ Can be a written or verbal description, a visual display, a 1. Ask - the analysts needed to define mathematical formula, or a what the project would look like and spreadsheet representation what would qualify as a successful result. These were the kinds of Three Forms of a Model questions they asked: The sales of a new product, such as a first- 2. Solid Preparation - The group built generation iPad or 3D television, often a timeline of three months and follow a common pattern. decided how they wanted to relay their progress to interested parties. ● Sales might grow at an increasing 3. Process - The group sent the rate over time as positive customer survey out. The data analysts also feedback spreads. (See the S- made sure employees understood shaped curve on the following slide.) how their data would be collected, ● A mathematical model of the S- stored, managed, and protected. curve can be identified; for example, S = aebect, where S is sales, t is time, e is the base of natural logarithms, and a, b, and c are constants. Predictive Decision Models ★ What will happen? DECISION MODELS ★ Why will it happen? ★ Accurate projections of the future A decision model is a model used to states and conditions understand, analyze, or facilitate decision- ➔ Predictive Decision Models often making. incorporate uncertainty to help managers analyze risk. Types of model input ➔ Aim to predict what will happen in ❖ data the future. ❖ uncontrollable variables ➔ Uncertainty is imperfect knowledge ❖ decision variables (controllable) of what will happen in the future. ➔ Risk is associated with the Types of model output consequences of what actually ❖ performance measures happens. ❖ behavioral measures
A Sales-Promotion Model Prescriptive Decision Models
In the grocery industry, managers typically ★ What should I do? need to know how best to use pricing, ★ Why should I do it? coupons, and advertising strategies to ★ Best possible business decisions influence sales. and transactions ➔ Prescriptive Decision Models help Using Business Analytics, a grocer can decision-makers identify the best develop a model that predicts sales using solution. price, coupons, and advertising. ➔ Optimization - finding values of decision variables that minimize (or Descriptive Decision Models maximize) something such as cost ★ What happened? (or profit). ★ What is happening? ➔ Objective function - the equation ★ Well defined business problems that minimizes (or maximizes) the and opportunities quantity of interest. ➔ Simply tell “what is” and describe ➔ Constraints - limitations or relationships restrictions. ➔ Do not tell managers what to do ➔ Optimal solution - values of the ◆ Mathematical models for decision variables at the minimum Total Cost (or maximum) point. ◆ A Break-even Decision Model A Pricing Model ◆ A Linear Prediction Model A firm wishes to determine the best pricing ◆ A Nonlinear Prediction Model for one of its products in order to maximize revenue. - - external groups are affected - - problem owner and problem Analysts determined the following model: solver are not the same Sales = -2.9485(price) + 3240.9 person Total revenue = (price)(sales) - - time constraints exist 3. Structuring the Problem Identify the price that maximizes total - Stating goals and objectives revenue, subject to any constraints that - Characterizing the possible might exist. decisions - Identifying any constraints or ❖ Deterministic prescriptive models restrictions have inputs that are known with 4. Analyzing the Problem certainty. - Identifying and applying ❖ Stochastic prescriptive models have appropriate Business one or more inputs that are not Analytics techniques known with certainty. - Typically involves ❖ Algorithms are systematic experimentation, statistical procedures used to find optimal analysis, or a solution solutions to decision models. process ❖ Search algorithms are used for complex problems to find a good - Much of this course is solution without guaranteeing an devoted to learning BA optimal solution. techniques for use in Step 4.
PROBLEM-SOLVING AND DECISION 5. Interpreting Results and Making a
MAKING Decision - Managers interpret the 1. Recognizing the Problem results from the analysis - Problems exist when there is phase. a gap between what is - Incorporate subjective happening and what we think judgment as needed. should be happening. - Understand limitations and - For example, costs are too model assumptions. high compared with - Make a decision utilizing the competitors. above information. 2. Defining the Problem 6. Implementing the Solution - Clearly defining the problem - Translate the results of the is not a trivial task. model back to the real world. - Complexity increases when - Make the solution work in the the following occur: organization by providing - - large number of courses of adequate training and action resources. - - several competing objectives DESCRIPTIVE ANALYTICS - “Analytics are a subset of … ★ Key task: Data access/shaping – business intelligence: a set of Power Query does this technologies and processes that use ★ Excel + Power Pivot data model data to understand business holds Past Business Results performance … The questions that ★ Pivot charts, Power View, Power BI analytics can answer represent the for data visualization higher-value and more proactive ★ Formulas: Sum, Count, Average, end of this spectrum.” – Tom Min, Max, Var, StdDev Davenport, Competing on Analytics FOUR CATEGORIES OF ANALYTICS ANALYTICS: THE THREE LEVELS ❖ Descriptive Analytics - The ● Descriptive Analytics: Classic Bl backbone of business intelligence, ○ Quantitative Assessment of descriptive analytics answers the Past Business Results fundamental questions of when, ○ Statistics, Exploratory Data where, what and how many. It Analysis, Visualization provides information about your ● Predictive Analytics customers and your business to help ○ Quantitative Methods to measure KPIs. Predict New Outcomes ❖ Diagnostic Analytics - Diagnostic ○ Forecasting, Prediction, analytics look at what already Classification, Association happened to find out why something ● Prescriptive Analytics occurred. It digs into the data to ○ Quantitative Methods to better explain anomalies and find Make Better Decisions hidden causal relationships. ○ Decision Trees, Monte Carlo ❖ Predictive Analytics - Perhaps the Simulation, Optimization most common of the four, predictive analytics understands correlations, CAN THIS HELP IN YOUR WORK? trends, and causation. It finds out ➔ Optimization models can deliver what will happen in the future. huge cost savings ❖ Prescriptive Analytics - Taking ➔ Simulation/risk analysis models can predictive analysis, a step further is help avoid disaster prescriptive analytics, which informs ➔ But very few business analysts have a business about what actions to the skills to do this take. With the help of artificial ➔ If you can do this, your value to your intelligence (AI), it tests variables to company will rise. Some analytic provide the best course of action models address operations, others before a company commits to address strategic decisions making a decision. ◆ Ex. whether to build a new ❖ plant, and where to locate it ➔ Be prepared to present your work to senior management What is Descriptive Analytics EXCEL ➢ Descriptive analytics is the process ➔ Data mining describes the next step of parsing historical data to better of the analysis and involves a understand the changes that have search of the data to identify occurred in a business. patterns and meaning. Identified ➢ Using a range of historical data and patterns are analyzed to discover benchmarking, decision-makers the specific ways that learners obtain a holistic view of performance interact with the learning content and and trends on which to base within the learning environment. business strategy. ➢ Descriptive analytics can help to EXAMPLES identify the areas of strength and ● Summarizing past events such as weakness in an organization. regional sales, customer attrition, or ➢ Descriptive analytics is now being success of marketing campaigns. used in conjunction with newer ● Tabulation of social metrics such as analytics, such as predictive and Facebook likes, Tweets, or ollowers. prescriptive analytics. ● Reporting of general trends like hot ➢ In its simplest form, descriptive travel destinations or news trends. analytics answers the question, "What happened?" Examples of descriptive analytics exist in ➢ Descriptive Analytics, the every aspect of the business, from finance conventional form of Business to production and sales, including the Intelligence and data analysis, seeks following. to provide a depiction or “summary ● Business reports of revenue and view” of facts and figures in an expenses, cash flow, accounts understandable format, to either receivable and accounts payable, inform or prepare data for further inventory and production. analysis. ● Financial metric and other business ➢ It uses two primary techniques, KPIs are examples of descriptive namely data aggregation and data analytics. These include metrics that mining to report past events. It assess the health and value of a presents past data in an easily business, such as the price to digestible format for the benefit of a earnings ratio, current ratio and wide business audience. return on invested capital. ● Social media engagement: How does descriptive analytics work? Descriptive analytics generates ➔ Data aggregation and data mining metrics that help determine the are two techniques used in return on social media initiatives, descriptive analytics to discover such as growth in followers, historical data. engagement rates and revenue ➔ Data is first gathered and sorted by attributable to specific social media data aggregation in order to make platforms. the datasets more manageable for ● Surveys: Descriptive analytics analysts. produces summaries of internal and external survey results, such as a ➔ Descriptive analytics also helps net promoter score. businesses communicate information among departments and 5 Steps in Descriptive Analytics to people outside the company. 1. Business Metrics and KPIs ➔ Potential lenders and investors, for 2. Data Collection and Aggregation example, may want to scrutinize 3. Data Extraction revenue, profit, cash flow, and debt 4. Data Analysis metrics before they put money into a 5. Data Presentation business.
How is Descriptive Analytics Used? What does it Tell Us?
● Reports: The key financial metrics ★ Current business performance included in a company’s financial ★ Historical trends statements are generated by ★ Strengths and weaknesses descriptive analytics. Other common reports also use descriptive analytics DRAWBACKS OF DESCRIPTIVE to highlight aspects of business ANALYTICS performance. - The primary drawback of descriptive ● Visualizations: Displaying metrics analytics is that it simply reports in charts and other graphic what has happened, without representations can more efficiently exploring the causes or attempting communicate their impact to a wider to predict what will happen next. It’s audience. also generally limited to relatively ● Dashboards: Executives, simple analyses that examine the managers, and other employees relationships between two or three may use dashboards to track variables. progress and manage their daily workload. Dashboards present a TYPES selection of KPIs and other ❖ Measures of Frequency: important information tailored to the ➢ Count, Percent, Frequency needs of each person. ➢ Shows how often something occurs Why is it important? ➢ Use this when you want to ➔ Descriptive analytics helps everyone show how often a response in the company make more informed is given decisions that guide the business in ❖ Measures of Central Tendency the right direction. ➢ Mean, Median, and Mode ➔ It reveals patterns that might ➢ Locates the distribution by otherwise be hidden in raw data, various points enabling managers to see at a ➢ Use this when you want to glance how well the business is show how an average or performing and where most commonly indicated improvements may be needed. response ❖ Measures of Dispersion or • Data and technology should be made Variation available at every corner of an enterprise to ➢ Range, Variance, Standard develop and nurture a widespread data- Deviation driven culture. ➢ Identifies the spread of scores by stating intervals • If data and analytics are aligned with ➢ Range = High/Low points overall business goals, then day-to-day ➢ Variance or Standard business decisions will be more driven by Deviation = difference data-driven insights. between observed score and mean • As people drive businesses, the ➢ Use this when you want to manpower engaged in Data Analytics must show how "spread out" the be competent and adequately trained to data are. It is helpful to know support enterprise goals. when your data are so spread out that it affects the • A centrally managed team must lead the mean analytics production and consumption ❖ Measures of Position efforts in the enterprise to bring behavioral ➢ Percentile Ranks, Quartile change towards a data culture. Ranks ➢ Describes how scores fall in • The concept of Data Analytics must be relation to one another. spread through both formal data centers Relies on standardized and informal social networks for an inclusive scores growth. ❖ Use this when you need to compare scores to a normalized score (e.g., a national norm)
According to a recent Forbes study titled
EY-Forbes-Insights: Data and Analytics Impact Index“people and culture” can influence the intelligence gathered from Business Analytics.
This study conducted jointly by
Forbes Insights and EY interviewed global executives and concluded that:
• Every modern business needs to build its
Data Analytics framework, where the latest data technologies like Big Data play a crucial role.