Our webinar co-hosts Kelle O’Neal and John Ladley kicked off the call with an explanation of why we use statistical analyses. Simply put, it’s because we have a question or problem to figure out that we think data can solve. We may find ourselves with too much data and too little time to analyze all of it — or we may not have enough resources to collect all available data for analysis. And that’s where data analytics comes in.
Kelle and John covered three types of analysis on the Data Insights & Analytics webinar:
- Descriptive – Aims to help uncover valuable insight from the data being analyzed and answers the question “What happened?” There are two overarching types of descriptive analyses. The first is measures of central tendency, and the second is measures of dispersion.
- Predictive – Helps predict unknown data using known data and answers the question “What could happen?” Common predictive analytics models include forecasting, simulation, regression, classification and clustering.
- Prescriptive – Suggests conclusions or actions that may be taken based on the analysis and answers the question “What should be done?” This can be done using a variety of techniques, including: linear programming, integer programming, mixed integer programming and nonlinear programming.
If you’re new to the topic of data analytics, like many of the attendees (based on our online poll at the start of the webinar), how do you know which type of analysis to pursue? Kelle and John suggested considering these factors:
- How much time you have
- The resources available to you
- The accuracy of your data and how accurate you need the model/analysis to be
- Your organization’s acceptance of the model you are considering
At the end of our March webinar, John closed by saying that statistical analysis isn’t a replacement for your own logic (“Don’t go on ‘statistical autopilot’.”) He reminded the attendees that Big Data isn’t a requirement for most analyses, as data analysis has been done for years.
John highlighted three sources for learning more about data analytics:
- Not So Standard Deviations podcast, “the latest in data science and data analysis in academia and industry”
- Tech Target’s When Predictive Models Fail, part of its Business Information publication
- Statistics in Plain English, a book by Santa Clara University professor Timothy C. Urdan
If you’d like to hear the full replay of our March webinar, visit DATAVERSITY’S on demand archive.
We hope you join us for next month’s DIA webinar on April 6, Building a Flexible and Scalable Analytics Architecture (learn more and RSVP).
