In this video, the instructor demonstrates gathering financial data, and shows how to gather financial data from the Federal Reserve Economic Database (FRED).
- [Instructor] All right, we figured out our question. We're going to forecast a key variable that our company has never thought about before. Now we need to gather the data that's going to let us do that forecasting. The reality is that your firm and all businesses out there have access to lots of data. You might not realize it though. You have proprietary data, like customer data that's available to you, but you also have publicly available data. For instance, from the US Census Bureau and the Federal Reserve. The data that your going to need depends on the question that is being asked.
Typically, a mix of public and private data is going to be what's needed in most circumstances. Where can you get data for a question you might be interested in answering? Well, there's three different techniques. First, you can buy it, you can build it based on your own proprietary data, or you can gather it for free. There's lots of the data vendors out there that will sell you data. Just do a Google search and you'll find many of them. I can't really talk about the proprietary data that you'll have available. That depends on the particular company.
But I do want to talk to you about the data that you can gather for free online. Before we get to that though, think about what kind of data you need to be looking for. What your data needs are going to be driven by your project needs. In particular, think about what it is you're trying to model Build a model and then get your data. If you're trying to optimize pricing, for example, you need very different data than if you're trying to identify employee theft problems.
Figure out what your model is, what your regression is going to look like, and then go find the data to populate that regression. In essence, we're asking, "What are the driving factors "that are going to influence the outcome that we care about?" There's a number of major public sources of data that are available, The Bureau of Labor Statistics, the Federal Reserve, and the US Census Bureau, for example. Google Trends is also great for survey data. And further, in today's day and age, with Facebook and LinkedIn, and a variety of other tools out there, you can actually gather customer comments with social media data online, although this can require textual analysis in some cases.
Let's take a look at how we can gather data from say the US Census Bureau though. So once we've pulled up Google, we're going to search for what's called Data ferret. DataFerrett is a tool from the US Census Bureau that lets us gather data across a wide variety of variables that the US Census Bureau has available. In fact, we can gather data not only on things like how many customers live in a particular area, but what their household income is, what their level of education is, et cetera.
This is the DataFerrett website. We can get started using DataFerrett by unblocking our pop-up windows. Once we do that, we have a variety of different data sets that are available to us. The American Community Survey is one of my favorites, for example. The American Community Survey has data on education, marital status, where our particular customers are born, and it covers more than three and a half million different households across the country on a regular basis, so we can use it to build census block by census block characteristics of our particular customers.
That census block is really akin to a neighborhood. This is a really powerful tool for actually going through and getting data. Once we're ready to begin gathering data, we'll click Launch DataFerrett and this will bring up a variety of different tools that we might want to look at. Now I'm not going to take you through all of the different tools that are out there, all the different data that's out there because there are so many different topics. But, for example, we could look at economic indicators based on what's available to us, and we have a variety of different kinds of publicly available data.
In this case, everything from monthly wholesale trade to new residential sales that is home sales, construction, sales in various industries, like food services, et cetera. I'd recommend that you go and check out some of this data that's available and see if you might be able to use it for your particular needs. In addition to data from the US Census Bureau, there's also data available from the Bureau of Labor Statistics and the Federal Reserve. Now, if you need specific data, buying it might realistically be the only option.
For example, financial data is often available online, say, price to earnings ratios for various companies or earnings reports. But in order to gather it in mass, you're typically going to need a Python script and so buying that kind of data is usually the most realistic option. Building data sets from your own data is crucial. The most interesting questions for a company can be answered using their own proprietary data. You want to look at opportunities to tap customer databases that you have and potentially develop in-house methods of collecting additional data.
Even think about surveying your customers. That can be a great option too. For example, I recently worked on a survey with an investment banking group. They were trying to predict the characteristics of particular customers that let them win deals. They were trying to figure out what kind of characteristics they should look for in prospective customers and prospective deals in the future so that they could focus their sales efforts on those opportunities. This is the kind of thing that you can do on your own as well.
Finally, you might think about doing surveys on other people's customers or OPCs. There's a lot of tools out there for this. For example, CriticalMix and SurveyMonkey are just two of the many tools that are available. The key you have to remember with any of these kinds of surveys is what's called survey data bias. In particular, if you're trying to figure out how to sell to other customers and you're surveying your own customers, you have to ask yourself, "Are my own customers representative "of the rest of the world "or am I gathering data on the people "that already like me and maybe I'm missing out "on the factors that caused other people "to not be willing to do business with me?" Keep these considerations in mind going forward, especially when you're using survey data.
Join Professor Michael McDonald and discover how to use predictive analytics to forecast key performance indicators of interest, such as quarterly sales, projected cash flow, or even optimized product pricing. All you need is Microsoft Excel. Michael uses the built-in formulas, functions, and calculations to perform regression analysis, calculate confidence intervals, and stress test your results. You'll walk away from the course able to immediately begin creating forecasts for your own business needs.
- Understanding big data and predictive analytics
- Gathering financial data
- Cleaning up your data
- Calculating key financial metrics
- Using regression analysis for business-specific forecasts
- Performing scenario analysis
- Calculating confidence intervals
- Stress testing