One question we frequently receive is how do you go about setting up an analytics problem? This is a complex question, so we will explore this concept generally. In future blog posts, we will explore each of these steps in greater detail.

Setting the Context

The first step to building an analytics model is to frame the context clearly. Some leading questions that may help are:

1. Who is the stakeholder (or end user)?

2. Can you clearly state the goal in one or two sentences?

3. By completing this analysis, what will the outcome be?

As an example, let’s say you are a risk manager for a commercial real estate investment firm. You might be tasked with determining the governance risks associated with development in a particular city or zip code. Your end stakeholder may be the investor. Your goal is to quantify the risk for the investor by identifying key governance indicators. The outcome we want is a specific measure of governance risk for a particular asset or location. Since governance can encompass many different measures, analysts have some leeway in choosing an approach. In the future, we expect ESG approaches to narrow down as standardized approaches become more commonplace. In the example above, the analyst may choose local taxation history and taxation outlook as possible measures of sustainable risk in governance.

Sourcing Your Data

This will be highly dependent on your question, and the outcome of the data selection aspect is largely affected by the experience of the analyst. We realize this isn’t the most helpful, but data selection choices absolutely make the difference between a passable and a superb analysis. Related to the above example, a passable approach would look at the municipal commercial tax rate. A superb approach may (1) look at historical trends, (2) identify key current and future proposed ordinance changes, (3) account for related municipal growth factors, or even (4) measure the local opinion about taxation now and in the future. As a starting point, we recommend taking your context statement from the exercise above and brainstorming what types of data would be ideal for solving that problem. Searching online can often net some rapid and free data.

Selecting Analysis Methods and Techniques

If you thought finding and selecting the right data was hard, choosing the right analytics method can be much harder. Statisticians, Data Scientists, and Engineers make careers dedicated solely to this step in the process. In our team, we have many full time Data Scientists dedicated to exactly this. With that said, this step can be approachable to many people who may not have as much technical training. We recommend focusing on selecting simple data sets and a simple analysis that can be accomplished in excel. Since the approaches and methods are quite varied, we will follow up with a more detailed exploration about how to select methods and techniques in a subsequent post.

Finding Insights

Finding insights is in many ways the fun part of the analysis. Once the analysis methods have been executed, it’s a matter of re-evaluating the context statement in terms of the analytics output.

Example

IS A MUNICIPAL BOND A WISE INVESTMENT?

See the below example for how we have framed a risk analysis rating on municipal bond investments.