Decision making happens at every level in an organization. Most folks understand bar charts and spreadsheets, but they might not comprehend the truly transformational power of advanced analytics that is now more widely available to everyone. Advances in modern visual analytics technology provides unprecedented, easy access to sophisticated algorithms. Advanced analytics techniques data scientists, statisticians and mathematics pros use are now literally button-click activities.
To get the most value from these innovations, it is important to have a fundamental understanding of common advanced analytics techniques, how they work, what data to provide and when to use them. In this article, I will share a quick overview with you. I also invite you to attend an upcoming webinar for a more detailed walk-through of specific techniques and use cases with Pyramid Analytics.
Why Use Advanced Analytics
First of all, the problems business face today can be complex. Multiple types of advanced analytics are often used to evaluate potential courses of action. Secondly, decisions today often need to be made faster than humans are capable of manually analyzing data. Advanced analytics can help with both of those needs and more.
- Can help you save money, make money or improve processes
- Provides deeper insights – answers what, why, and presents potential outcomes
- Makes predictions or generates actionable recommendations to improve outcomes
- Empowers repeatable or automated decision-making processes
Advanced analytics is literally the art of the possible. It allows you to intelligently drive your organization further, forward, faster to achieve better outcomes.
From Insights to Outcomes
Descriptive analytics provides a quantitative assessment and visualization of past results. Typical activities include exploratory data analysis, applying general statistics and outlier detection. Often data analysts will use descriptive analytics to get a better understanding of the available data and overall context. They will look at data shape, location, proximity, variability, relationships and patterns.
Where descriptive analytics is reactive in nature and allows an understanding of what has happened in the past, advanced analytics using predictive and prescriptive analytics to support more proactive optimization of what is best in the future based on a variety of scenarios.
Predictive analytics is an area of statistics that is focused around capturing relationships between explanatory variables and predicted variables from past occurrences and using them for prediction. Often predictive analytics involves data mining – automatically discovering interesting patterns in data. Common predictive analytics activities include forecasting, classification, and association.
Don’t assume predictive is always future-oriented. Predictive analytics can be applied to the past, present or future. The accuracy and usability of your models varies upon the level of analysis and quality of your assumptions. We are already seeing more predictive and artificial intelligence being embedded into reports, business processes and applications.
Prescriptive analytics provides the best options for given situations based on the concepts of optimization. It lies at the far end of the analytics maturity spectrum. Prescriptive modeling and optimization is an area of management science that has been historically referred to as operations research or decision science. As data-driven organizations continue to recognize that information is a strategic competitive advantage, they will strive towards using more prescriptive analytics due to the actionable, recommendations that it delivers.
Although prescriptive analytics has exceptionally high business impact potential, it can become overwhelming and complex rather quickly. As a result, this area of analytics is often an untapped, truly golden window of opportunity to explore in most organizations.
Common prescriptive analytics include business decision what-if and goal seeking models, scenarios, optimizations and simulations. These techniques are run with known and randomized variables to gain a better understanding of a range of possible outcomes.
Where to Start
One of the most common questions that I get when teaching advanced analytics is where should we start? Ideally you would identify a business problem that can be solved with a straight forward to understand algorithm such as a classification model or decision tree. This allows you to learn the tools and be able to explain the output to use it.
Business analysts or technical practitioners should partner with subject matter experts in the business to validate and also improve your models. The joint team can review and apply findings to drive better outcomes that provide the biggest return.
I like to share a real story of building a model to predict insurance agent churn. Our first round of results suggested not to hire anyone from New York. When we consulted the subject matter expert, they added features to our model to include legal climate. Apparently a law had changed in New York that was skewing the results of our first model.
For a Deeper Dive
If you want to learn more about advanced analytics techniques and see how to use these features, please join me in the upcoming webinar next week. An on-demand recording will also be made available to you by signing up.
Specifically, we will explore how to apply proactive, predictive intelligence to enhance reporting. I’ll be walking through time intelligence, statistics, forecasting, clustering, R integration, and creative use of parameters and variables for identifying patterns and relationships in data that weren’t initially evident. I’ll also provide a rapid introduction to the most popular analytics languages today, R.
- Descriptive and Diagnostic: Data Exploration, Visualization, Chart Statistics, Pareto Analysis and Outlier Analysis
- Predictive: Regression, Forecasting, Clustering, Decision Trees, Neural Networks and customizing models with R Scripts
- Prescriptive: Scenario Variables and What-If Analysis and R Scripts
- Getting Started with R
Best of all, no statistics degree is required. Since visual analytics offerings like Pyramid Analytics now include simple point-and-click advanced analytics, anyone can start using it.
Good Reads and Resources
Additional resources to check out or bookmark for future reading are listed below.
- Pyramid Analytics
- Pyramid Analytics R Integration
- Open Source R Project
- R Studio
- Getting Started with R
- Additional R articles
- Practical Predictive Analytics
- Prescriptive Analytics Channel