About Data Science
Here at JBC Analytics, we want to help to demystify some of the concepts of Data Science and Predictive Analytics as these techniques are at the core the Analytics frenzy that is dominating the growth strategies of all enterprise level organizations. While Data Science techniques were first introduced around 1821 in the Scientific world, the use of these techniques in the ordinary course of everyday businesses is relatively new (less than 5 years).
Per Kirk Borne, PhD, Principal Data Scientist at Booz-Allen-Hamilton, Data Science is not an IT function. It is a core business function where big data is manipulated to identify new patterns, trends and anomalies to gain insights. These insights lead to discovering new markets, improving business, improving bottom line and producing new business outcomes using a variety of complex predictive and prescriptive analytical methods.
According to Salesforce, a leader in Consumer Analytics, “Predictive marketing uses data science to accurately predict which marketing actions and strategies are the most likely to succeed. In short, predictive intelligence drives marketing decisions.” A good example of this use would be the way Amazon, Netflix and others are able tomake accurate suggestions of what you like. This is accomplished using forms of Predictive Analytics, Machine Learning and Artificial Intelligence.
In short, because the collection of data is everywhere, the actions you’ve performedin the past allow your future wants and actions to be “predicted”. In technical termsthese strategies are variations of a statistical concept called Regression Analysis. Basic Regression Analysis is executed using a mathematical formula (built by a DataScientist) more commonly referred to as a model. Machine Learning and Artificial Intelligence are represented by groups of more complex models linked and/or nested together in complex Decision Trees. The “magic” of these techniques occurswhen the outputs of the models are then fed back into the overall algorithm, so it “learns” or “teaches itself” how to be more accurate for future predictions.
While there are thousands of predictive model types, AgilOne, a silicon-based leaderin the industry, says there are 3 primary classes of predictive models that apply wellin the Marketing domain.
- Cluster models (segments) – Used for customer segmentation; algorithms segment target groups based on numerous variables, everything from demographics to average order total. Common cluster models include behavioral clustering, product-based clustering (also called category-based clustering), and brand-based clustering. Can Respond Authentically
- Propensity models (predictions) – Used for giving “true” predictions about customer behavior. Common models include predictive lifetime value; likelihood of engagement; propensity to unsubscribe; propensity to convert; propensity to buy; and propensity to churn.
- Collaborative filtering (recommendations) – Used for recommending products,services, and advertisements to customers based on a variety of variables,
including past buying behavior. Common models (like those used by Amazon and Netflix) include up-sell, cross-sell, and next-sell recommendations.