Understanding the Data Analytics Spectrum
If you want to discover meaningful insights and make informed decisions about your work, projects, or even your personal habits, it is imperative that you understand the different types of analysis. There are four main types of data analytics, namely descriptive, diagnostic, predictive, and prescriptive analytics. Each one is used to accomplish a different goal and employ the use of varying techniques to help you make better decisions.
Descriptive Analytics
Descriptive analytics involves the examination of historical data to understand what has happened in the past. It answers questions like "What happened?" and provides a summary of key metrics and trends. For example, you would use descriptive analytics to determine whether young people were more engaged with your marketing content, what percentage of your students received As or Bs or how patients with other medical conditions responded to treatment when compared to those with no prior history. This type of analysis can be performed by compiling data in spreadsheets or in other formats and visualizing the data in graphs or charts that enable you to see distinct trends. Descriptive analytics is the foundation for further analysis as you can use your findings from descriptive analytics to determine identify patterns in the data, then use it as context for deeper analyses.
2. Diagnostic Analytics
Diagnostic analytics delves into the reasons behind past events in an attempt to answer questions such as “Why did this happen?”. Thus, it goes one step further than descriptive analytics and attempts to uncover the root causes of trends or anomalies identified in descriptive analysis. Diagnostic analytics helps organizations and people to understand the factors influencing their performance and enables them to address underlying issues. For example, if you’re a teacher, do you see any relationships between students’ performance and their attendance? If you’re a banker, do you see any common activities or characteristics between customers who churn? When a doctor gives you a diagnosis, it serves as an explanation for why you have been experiencing certain symptoms. This is exactly what diagnostic analytics does.
3. Predictive Analytics
Predictive analytics forecasts future outcomes by using historical data and statistical algorithms. There are a number of statistical models that you can use to perform predictive analytics such as linear regression, time series forecasting and decision trees. Being able to predict the future based on things that have happened in the past is a wonderful feat when done correctly. This is more advanced than the prior types of analytics as models are built to predict trends, patterns, or behaviors. You may notice a difference between certain groups in your data, but is the difference statistically significant enough to have a true impact? Statistical models would help you to determine this. While businesses typically have their own engineers building and customizing predictive models, there are also some apps and software that can do these predictions for you if you provide the data. For example, Weka is a free app that you can download and use to run models, and SAS also creates various platforms that do the same. Predictive analytics empowers organizations and people to anticipate potential outcomes and make proactive decisions to mitigate risks (especially financial risks or the risk of a store running out of stock), and it enables us to capitalize on potential opportunities.
4. Prescriptive Analytics
Prescriptive analytics takes predictive analysis to the next level by recommending actions to optimize future outcomes. Once you predict what is likely to happen, prescriptive analytics enables you to suggest the best course of action to achieve your desired results. An insight that is not actionable is just a fact, so with this type of analysis, the goal is to answer questions such as "What should we do?". Discovering actionable insights drive informed decision-making. For example, if you ran multiple marketing campaigns at different times over the course of multiple years and you discovered that the campaigns received the most engagement during specific timeframes, you can run a model to determine the factors that would make the campaigns successful and acquire useful information that will assist you with planning ahead for the next year.
Each type of data analytics plays a distinct role in the analytical process, offering increasingly sophisticated levels of insight and value. If you understand the spectrum of data analytics, you can empower your organization and yourself to turn data into insights, drive innovation, and yield competitive advantage in today's data-driven world.
