What is Predictive Data Analytics?
Predictive data analytics uses current and past data to let you make predictions about the future or other unknowns. You can see the likelihood of a coming event or a specific situation, given the data being analyzed. Predictive data analytics examples are wide-ranging:
- Ecommerce sites use predictive analytics to offer specific products likely to interest a visitor. Predictions are based on that visitor’s past purchases and viewing of products.
- A human resources department might use predictive analytics to detect if employees are thinking of quitting, and then persuade them to stay.
- In IT security, the prediction could be about where malware has infected systems, based on network activity and data flows. These systems then get top priority for in-depth inspection.
See e-commerce analytics in action:
See it in action:
Predictive data analytics differs from general forecasting. It gives you insights into individual cases (individual customers, employees, and systems, in the examples above). This makes predictive analytics actionable. It opens the door to immediate improvements and results by applying the insights from the analytics.
Bringing in Big Data
Big data and predictive analytics often go together. The richness of big data, can be leveraged for the highly specific insights per visitor. An example of such big data is the individual clicks on different products and pages of each visitor on an ecommerce site. Analytics techniques must then be adapted to high volume, velocity, and variety of data. One technique is data mining to pick out patterns. Others are statistical algorithms to build models, and machine learning to update models as new data arrives.
Basic Steps for Predictive Data Analysis
Making use of predictive data analytics can be done in these steps:
- Define the result you want, e.g. how to offer each customer additional products of interest.
- Collect the data that will be needed (perhaps ecommerce site tracking data, CRM logs, etc.).
- As necessary, prepare the data from each source, then combine the different datasets.
- Make predictive analytics models, using statistical analysis to see which outcomes typically follow which events.
- Apply your models to your business.
- Review the models to ensure they are working properly.
User-friendly analytics software can make these steps accessible to business and non-technical users. You still need to decide which business benefit you want and identify the data required. After, the right software application can help make data preparation and combination simple, and the construction of predictive analytics models intuitive.