Big Data and Business

The presentation form Srijan Kumar (Stanford)  gave insights on the components of Big Data and data mining. Describing Big Data as the combination of:

  • Machine Learning
  • Statistics
  • Artificial Intelligence
  • Databases

is a great way to look at it. I would like to add the notion that we can think of Artificial Intelligence as the combination of programming and statistics. An important addition to this view are the multiple layers of Big Data systems:

  • Data Visualization
  • Data Analysis
  • Data Structure
  • Data Storage

Usually, the first step in processing large data sets, especially in data visualization is data cleansing. That means you would always want to clean your data before you will start visualizing them with tools like Tableau ( ), Pentaho/Hitachi Vantara ( ), Tibco / Jaspersoft ( ) or similar tools from other vendors.

IBM data scientists break Big Data into four dimensions, or “The Four V’s of Big Data”:

  1. Volume: The Scale of Data
  2. Variety: Diversity of Data
  3. Velocity: Speed of Data
  4. Veracity: Certainty of Data

IBM adds to this fifth “V” as Value, meaning the ability to achieve greater value through insights from superior analytics ( ). Applications of Big Data are solving real-world problems. The presentation gave as examples recommender systems, market basket analysis, spam detection or duplicate document detection.

Let’s have a look how addressable markets of Big Data applications for enterprises could be described and monetized:

  1. Advertising & marketing analytics from user data
  2. Consumer lifestyle and segmentation services for marketing campaigns
  3. IoT services through data derived from sensors for optimization tasks
  4. Profitability improvement for financial calculations and transactions

For example in advertising user-related data are used to identify and monetize audiences. That means this user-related data delivers actionable insights once it is processed. Predictions on buying behavior can be made and help marketers to target desired audiences. In consumer lifestyle and segmentation services, Big Data and Artificial Intelligence (BDAI) opportunities are delivering lifestyle, demographic and credit-risk segmentation to support third-party marketing campaigns. The BDAI opportunity in the Internet of Things (IoT) is to convert data into meaningful analytics to improve crop yield, energy demand, smart city traffic optimization, and operational usage in health care. Profitability improvement means using data for financial calculations to improve customer acquisition, retention and average revenue per user (ARPU) or to reduce operational expenditures (OPEX), capital expenditures (CAPEX) and to optimize revenues.

The following article from recode gives a great overview of a McKinsey research report on how AI could create trillions worth of value ( ). Mc Kinsey focused on 19 industries in the report and business functions where AI could drive the most impact are marketing and sales and supply-chain management and manufacturing. According to the report, AI could improve personalized recommendations for e-commerce or predict traffic patterns and reduce trucking costs ( ). More interesting thoughts on what to expect from AI can be found in a techopedia article ( ). The article calls data analytics a $200B market. Even if we have to be careful with market numbers, the dimension of the number shows that it is expected that Big Data technologies will distill data into actions with business value.

Looking at the applications of AI in marketing the following article groups artificial intelligence techniques in marketing into four stages ( ):

  1. Reach – e.g. smart content curation, showing visitors content relevant to them
  2. Act – e.g. ad targeting, ads that perform best on specific users
  3. Convert – e.g. dynamic pricing, special offers that are necessary to convert the target
  4. Engage – e.g. marketing automation, determine best times and ways to interact with the user

As we learned, data mining is the extraction of information from very large data sets. For these large data sets and especially for personal data, it is very important to secure the data and to control the access to these data sets. The lecture gave highlights on the latest examples of cybersecurity breaches. The economic damage is huge. In the case of the Yahoo breach, it reduced the price tag by more than US$300M for example. Public companies display their business risks and opportunities in their annual reports. I believe we should expect that they will have to assess their risk of exposure to cyber attacks in the same fashion they describe their business environment.

Currently, the FTC and FCC regulate in certain ways how companies can monetize their users’ data. A lot of the great online services that consumers use are free because their data is monetized by the providers of these services. Let’s think about if we as consumers were exclusively monetizing our own data – would we benefit more from that and would all the great internet services have been built and be accessible to us? How would the services be priced in that case?


One comment on “Big Data and Business”

  1. This overview of the applications of big data to business is really eye-opening and now I can’t help but wonder what the market for analysts begins to look like as this business need develops. “Big data” analysis seems like it demands skills in programming and statistics (which command high salaries in the labor market). However, as you mention above, companies are already providing enterprise software applications that serve to simplify the process of analysis, so that such skills are not as deeply essential. This seems like a necessary step if businesses are to truly unlock the value of big-data on a widespread scale. Data is constantly being produced and all data is probably of value to someone, but will only be tapped if it is not prohibitively costly. What then becomes of the employees and would-be employees in this field? Does “Big Data analysis” only remain a widespread lucrative career option for a short while until the data processing applications take over, or will the role of the analyst simply adapt (perhaps to a higher level of abstraction) as needs change?


    Users who have LIKED this comment:

    • avatar

Comments are closed.