Why Predictive Analytics Has Become Essential For Optimized FP&A

The role of the FP&A professional continues to progress and evolve along a continuum. In the past, we spent most of our time and effort explaining what had happened and tried to divine the future based on our observations. At the time, the biggest challenges were a lack of data and tools powerful enough to permit extrapolating limited available data to project into the future.

Today, FP&A professionals focus largely on explaining what is occurring throughout their organizations in real time. By adopting philosophies like dynamic planning, we can apply real-time insight into our actuals and adjust business strategies on the fly to increase the probability of achieving stated goals.

What we truly aspire to become is that business partner/advisor who can provide valuable insights and foresight to our organizations. Organizations around the world find themselves at different points along this FP&A maturity curve.

Decision Support

Predictive analytics (PA) can help organizations better understand the most likely outcome of a certain business scenario and deploy the proper mix of resources (time, people, money) to maximize the benefit and/or minimize the cost. PA helps us convert data into information, and use that information to make better, faster, smarter business decisions.

PA uses data, statistical algorithms, and machine learning techniques to identify the likelihood of future outcomes based on historical data. The goal is to go beyond knowing what has happened in the past to provide the best assessment of what will happen in the future. Though PA has been around for decades, companies often fail to leverage the technology to its full potential. More and more organizations are turning to PA to increase their competitive advantage.

Why Now?

The amount of data available to us has never been greater, and continues it to grow by leaps and bounds. It wasn’t so long ago that our access to data was greatly constrained by time and money. Now, for all practical purposes, data is free, instant, and unlimited.

Technology has advanced to a point where we now have tools that are powerful enough to manipulate all this data to help us make better business decisions. Because of these advances in technology, PA is no longer just the domain of mathematicians and statisticians. Now FP&A professionals are utilizing these tools to beyond learning what happened and why to discovering insights about the future.

Predictive Versus Descriptive Models

PA models use known results to develop or train a model that can be used to predict values for different or new data. Modeling provides results in the form of predictions that represent a probability of the target variable based on the estimated significance from a set of input variables.

This is different from descriptive models that help you understand what happened, or diagnostic models that help you understand key relationships and determine why something happened.

There are two types of predictive models: classification and regression. Three of the most widely used predictive modeling techniques are decision trees, regression, and neural networks. Decision trees are classification models that partition data into subsets based on categories of input variables. Regression (linear and logistic) is one of the most popular methods in statistics used to determine relationships among variables. Neural networks are sophisticated techniques capable of modeling extremely complex relationships. They are popular because they are powerful and flexible.

Fun with Modeling

If you are a FP&A professional, like me, this is the kind of work that gets you up early in the morning, keeps you fired up all day, and makes it difficult to turn off your laptop in the evening.

In future articles, we will expand upon the modeling techniques available to today’s FP&A professionals. These techniques include Bayesian analysis, ensemble models, gradient boosting, incremental response, K-nearest neighbor, memory-based reasoning, partial least squares, principal component analysis, support vector machine, and time series data mining. It may just be the Quant in me (or the geek) that truly loves this kind of work.