Data analytics begins with data. You often have to collect data before you work with it and this can range from doing a survey to finding publicly-available data on the web. Then, once you have the data you often need to clean and organize it. Here is a breakdown of the data analytics process:
The Basic Steps in the Data Analytics Process
- Collect
- Clean
- Organize
- Analyze
- Communicate
Here you can see that the analysis is only one step in the process. Before the analysis, you need to collect, clean, and organize the data. Collecting the data can involve finding it on the internet or tracking it down internally within a firm. Excel has a plug-in for a data “feed” using an API. Cleaning the data validates that the data is accurate. You do not want to move on to the analysis until you know that you have data of sufficiently high quality. As the phrase goes, “Garbage in, garbage out.” Once you have completed the analysis, you then need to communicate the findings effectively, including data visualization.
The goal of data analytics is to simplify data and amplify its value. It’s power is to reduce large volumes of data into an actionable insight. What is the question? What is the key takeaway? This is a form of communication that goes beyond quantitative analysis. Part of learning business analytics is learning how to convert data into meaningful language for others. Analytics must be useful to the user or it will not be used.
In addition to the data analytics process, you should also understand the different types of data analytics. Data analytics typically is described as having four types:
The Four Types of Analytics
- Descriptive analytics
- Diagnostic analytics
- Predictive analytics
- Prescriptive analytics
It is helpful to distinguish these four categories because they build on one another. Descriptive analytics is usually the easiest and most direct, but it lays an important foundation for the other three categories. You can think of each category as answering a question. Descriptive analytics answers “What happened?; Diagnostic analytics answers “Why did it happen?”; Predictive analytics answers “What will happen?”; and Prescriptive analytics answers “What is the best course of action?”
Here is an external article on the four types of data analytics that includes a helpful visualization of how they work together and description of each one. Just as these different categories of data analytics can be useful for answering different types of questions, each of them requires a different type of quantitative strategy.
Descriptive analytics is always a good place to start because it helps you understand and describe the data that you have. It helps you understand “what” is there and then you can move on to the more interesting “so what?” Basic facts are the foundation of good decision-making. For instance, tracking revenues and expenses is essential before you can optimize for increased profitability.
Diagnostic analytics is useful for assessing business performance. Companies can use evaluative analytics to monitor trends in their own firm or in their industry. With this type of monitoring, companies can quickly address small problems before they become big ones.
Predictive analytics is about predicting future outcomes using historical data. Recommendation systems are an example of predictive analytics because they predict what the customer is likely to want next. Inventory systems also use predictive analytics to predict future product demand and the need for future inventory. Internally, predictive analytics can be used for employee retention. Knowing which employees are most likely to leave can help a company focus their retention efforts on these employees. All of these applications use predictive analytics to gauge the probability of a future outcome based on data from the past. Winston Churchill said “The farther backward you can look, the farther forward you are likely to see.” Predictive analytics is extremely valuable because it provides an opportunity for optimization. A prediction of the future based on the past is very useful information.
Prescriptive analytics is about recommending a course of action. It is the highest value form of analytics, because it helps the user of the analytics (an individual or company) to make decisions to improve future outcomes. It is not just about what is going to happen (predictive), but how choices can be made to influenced what is going to happen. Just like a doctor prescribes a treatment to help a patient get better, a data analyst can prescribe a course of action to improve future performance. In business, predictive analytics can be used to increase future revenue, decrease future costs, and ultimately increase future profits.
Business analytics should always have the goal of solving a specific business problem. What question are you trying to answer and how does the data help you answer that question? You can think of a decision as an answer to a question. Should the company enter a new market? Should it close down certain operations? These types of difficult questions require data to support the decision.
The holy grail of data analytics is identifying causal relationships. If you know that increasing X causes Y to increase as well, then you can use X as a lever to control Y. For instance, more marketing leads to more sales. That’s good to know, because a firm can choose to do more marketing. But without knowing the causal relationship, it’s difficult to know whether more marketing is worth it. Cause-and-effect linkages help decision-makers analyze how much the influencing metrics are affecting the influenced metrics. The distinction between correlation and causation is very important, because you cannot use X to influence Y if the relationship is simply a correlation.
Moving from descriptive to diagnostic to predictive to prescriptive analytics is a progression in skill and contribution. The better you become at each of them, the more you will be able to transform data into a valuable asset.