Descriptive and Predictive Data Analytics in Big Data
IME provides machine learning, artificial intelligence solutions, and related analytics techniques, to handle big data and operational data; helping organizations and decision makers to get the optimum solutions in several fields. Big Data Analytics enables enterprises to explore their big data to the most relevant information and analyze it to predict critical business decisions. There are several types of data analytics in big data. In this article, we will focus on two types: descriptive and predictive analytics, showing their place and characteristics compared to other types, followed by more details and examples from real applications.
Big Data is data whose scale, distribution, diversity, and/or timeliness require the use of new technical architectures and analytics to enable insights that unlock new sources of business value . Big Data Analytics involves collecting data from different sources, mange it in a way to be consumed by analysts, and finally deliver data products useful to the organization business. The process of converting large amounts of structures/unstructured raw data, retrieved from different sources into a useful data product for organizations is the core of Big Data Analytics.
Types Of Analytics Techniques
Figure 1. Types of analytics techniques (Gartner, 2017) 
Figure 1 shows different types of analytics techniques . It is clear that the descriptive analytics type concerns what happened, this type can be applied using visualization reporting and data mining. For further analysis, predictive analysis can be done, estimating what will happen. Predictive analysis can be done using simulation and regression analysis. Furthermore, monitoring patterns for forecasting can be done in this type of analytics, which leads to insights creation. Whereas prescriptive analysis and artificial intelligence deeply go through taking decisions and optimum actions.
Figure 2 Gartner’s analytics maturity model 
According to Gartner classification, as shown in Figure 2, the chain of analytics can move from descriptive, which is more informative class and less difficult to implement, to predictive analytics, finalizing with prescriptive analytics, that is more complex and related to optimum actions. Prescriptive analytics goes above and beyond the previous classes by predicting not one possible future, but instead “multiple futures” based on the decision-maker’s potential actions .
It is the simplest class of analytics. It enables the analyst to change data into smaller and useful bit of information , or in other words related to data reduction and digestion. Some reports recorded that about 80% of business analytics are descriptive; for example, social media data that are used by social analytics.
It can be considered as “summary view” of facts and figures in an understandable format to either inform or prepare data for further analysis . The output of descriptive analytics is prepared to be inputs for more advanced predictive or prescriptive analytics that deliver real-time insights for business decision making.
Examples of Descriptive Analytics
- Company reports that simply provide a historic review of an organization’s operations, sales, financials, customers, and stakeholders.
- Summarization of past events such as regional sales, customer attrition, or success of marketing campaigns.
- Tabulation of social metrics such as Facebook likes, Tweets, or followers.
- Reporting of general trends like hot travel destinations or news trends.
It is a mean to predict future events based on past behavior. Both statistics and data mining are used in predictive analysis. It may need large data sets to identify patterns and trends, build predictive models, and create visual representations.
Some real examples of predictive analytics 
- Banking: the credit score, where data about customer previous payments and his financial behavior are used to predict his future payments.
- Marketing: predictive analytics are used to predict some important marketing measures like:
- Next best offer or product recommendation: predicts what your customer is most likely to purchase next.
- Return on Investment (ROI): Measures the benefit from an investment.
- Quarterly or yearly sales forecasts: Predicts how much income your company is likely to generate.
- Customer lifetime value (CLTV): Predicts how much a customer will buy over time.
- Healthcare: predictive analytics are used to maintain some important data, that are used later in the prediction process, like:
- Patient wellness management records.
- Biometrics usage, genomic data and family history data,
- Financial records (e.g. costs, insurance reimbursements, revenue and supply chain),
- Patient satisfaction surveys.
- Cybersecurity: By observing network traffic and identify the normal and anonymous patterns, to detect if any intrusions in the network. This this information helps to predict and detect frauds, vulnerability and threats.
Tips for Making Good Predictions
- Get good data
- Use recent and fresh data
- Make sure your data is complete
- Use the suitable predictive model
- Dietrich, D., Heller, B., & Yang, B. (2015). EMC Education Services. Data science & big data analytics: Discovering, analyzing, visualizing and presenting data.