Join us on April 7-9, 2021

The first industry event focused on data engineering

Register Today
All Blog Posts

Data Mining and Predictive Analytics: From Insight To Action

August 9, 2016

Data mining and predictive analytics are a winning combination for business success.  But you can’t get to the latter without the former. You have to have data mining to find insights and relationships, but it’s predictive analytics that takes those insights and turns them into consistent, scalable action that improves your organization. Data mining is to predictive analytics what the swimming pool is to an Olympic diving hopeful: you can’t have the latter without the former.

Data Mining : The Currency of Predictive Analytics

You can’t pay your rent with gold still in the ground. Until you dig it out of the ground, it’s practically useless. Data mining, which leverages aggregate data gathered over a long time period and with multiple sources, is about uncovering relationships between the variables in that data. The implication is that data mining operates on aggregate data, and focuses on the correlation (“What happened?”), but not the causes and/or effects (“Why? And what was the impact?”)

For example, data mining might show a credit card company that shoppers who charge over $1000 in Rio de Janeiro are more likely to file a fraud claim vs when they spend the same amount in London. What data mining alone cannot do is tell you how to know at the moment of sale whether a business traveler is entertaining an important client at an expensive restaurant, or whether her wallet was stolen on the beach and being used by a thief.

Predictive Analytics: The Payoff Of Data Mining

On the other hand, predictive analytics surmises outcomes from measurable results, by using data patterns to create mathematical models formalizing relationships, and from those models, consistently and accurately predict future behavior.

For example, in the statement, “Anne’s credit card was used fraudulently in Rio De Janiero a few seconds ago, and there was a .05% fraud rate in Rio in August,” fraud is an outcome in this case—an example of data mining. But predictive analytics attempts to calculate the potential for fraud in real time, using geographical location, spend velocity, category spends, and other variables that relate to Anne, the card holder in question. Predictive analytics can then then compute an index based on which it will qualitatively flag the transaction as fraudulent for human review. Fortunately, Anne notified this card issuer of her upcoming travels to Rio, so they didn’t have to suspend her card and embarrass her in front of her big client at dinner.

Data Mining and Predictive Analytics Can Transform Your Organization

Data mining and predictive analytics can help you find clues about what happened and then show you what’s next. Trifacta was designed from the ground up to help reduce data cleaning and data preparation time for data mining and predictive analytics by enabling better assessment of data sources, offering smart extraction that learns preferences over time, and providing easy to use, intelligent, interactive, visual data analysis that improves data understanding for your entire organization.

To learn more about leveraging Trifacta to execute data mining and predictive analytics initiatives, read our white paper, “Best Practices for Executing New Analytics Initiatives”