Business Analytics and Big data

Summary

This lecture introduces business analytics and big data as tools for turning the growing volume of business data into better, fact-based decisions. It concludes the course’s overview of e-business technologies and leads into upcoming material on decision-support systems.

Business analytics

Data is produced by transactions and by sources such as products, locations, sensors, RFID devices, GPS, and social media. Data alone may be of little use; analytics applies information technology, statistics, quantitative methods, and mathematical or computer-based models to extract insights and support decisions.

The lecture divides analytics into three types:

• Descriptive analytics — What happened, what is happening, and what is the problem?

  • Examines and presents existing or current data.
  • Tools include ad hoc reports, dashboards, data warehouses, and alerts.
  • Helps define business problems and identify opportunities.

• Predictive analytics — Why might it be happening, and what may happen next?

  • Uses methods such as data mining, text and web mining, and forecasting.
  • Produces projections of future conditions to inform decisions.

• Prescriptive analytics — What should be done, and what is the best option?

  • Evaluates possible actions, including “what if” scenarios.
  • Uses optimization and related methods to recommend actions.

Retail markdown example: A store deciding when and how much to reduce seasonal prices can:

1. Use descriptive analytics to review historical prices, sales, and advertising for similar products.
2. Use predictive analytics to estimate sales under different prices.
3. Use prescriptive analytics to choose pricing and advertising strategies that could maximize sales revenue.

Analytics in supply chains

Supply-chain analytics depends on resources to collect and manage data from systems such as RFID, ERP, and other databases. It can combine data analysis and data mining with supply-chain planning. Statistical process control can also support performance management, including Six Sigma processes.

Big data

The lecture describes big data as data whose scale, diversity, and complexity require new architectures, techniques, algorithms, and analytics to manage it and uncover value. It notes that definitions and the number of “Vs” are not standardized.

The main characteristics discussed are:

• Volume: The amount of data is growing rapidly, particularly as connected devices and other sources continue generating data.
• Variety: Data arrives in many forms, including text, numbers, images, audio, video, sequences, and time series. Sources include RFID, GPS, sensors, customers, and social media. Data may be structured or unstructured, static or streaming.
• Velocity: Data can arrive continuously and rapidly—for example, from RFID or GPS devices and changing stock prices. Timely analysis can be important for responding to opportunities or problems.
• Veracity: Some accounts add this fourth “V,” referring to uncertainty, inconsistency, incompleteness, ambiguity, latency, and other data-quality issues.

The lecture also distinguishes between data at rest—stored data ranging from terabytes to exabytes—and data in motion, or streaming data that may require responses within milliseconds or seconds. It emphasizes that combining data from different sources is often necessary.

Sources, technologies, and applications

Big data may be generated by social media networks, scientific instruments, mobile devices, sensors, and networks. The lecture refers to transaction processing, analytical processing, and real-time analytics as relevant technologies. It also contrasts conventional business-intelligence reporting and data mining with predictive analytics that may use larger, more diverse, and more timely datasets.

In supply chains and related business functions, examples of data include:

• Sales: Price, quantity, item, time, date, and customer information; data may come from direct, distributor, internet, international, or competitor sales.
• Consumer behavior: Items browsed or bought, purchase frequency and value, click-through activity, credit-card use, product reviews, social-media activity, and other indications of preferences.
• Inventory: Product details such as style, color, and size, with updates from warehouses, stores, online stores, and vendors.
• Location and movement: Sensor data about where goods are, who moved them, their route, and potentially their future path.

Potential supply-chain applications include:

• Improving demand forecasts by connecting real-time sensor data with machine-learning methods.
• Responding more quickly to demand using checkout and RFID data.
• Designing warehouses and choosing locations using optimization and network analysis.
• Evaluating and selecting suppliers with more factors and up-to-date information.
• Assigning transportation resources, such as trucks and rail, in response to current conditions.

The lecture names UPS as an example of tracking packages and monitoring vehicles, and Schneider National as an example of using sensor locations and driver-behavior data. It also notes that big-data-driven decision-making is being used in grocery retail, airlines, and trucking.

Speakers and sources featured

• Speaker: One lecturer, who presents the material; no personal name is given in the subtitles.
• Sources or organizations mentioned: An unnamed external source is credited for a big-data definition and some facts. The subtitles also mention Walmart, UPS, Schneider National, and general sources of data such as RFID, GPS, sensors, social media, and scientific instruments. No other speakers are identified.