Descriptive Reports
Descriptive reports are the reports that represent what happened in the past using graphs and key parameters indicators (KPIs). They explain or describe why the event happened by performing a root cause analysis.
Operational reporting is a good example of descriptive reporting. It is used to analyze the performance of ongoing business operations and related short-term business trends and patterns from historical data. These reports are the main bread and butter of BI reporting for lower and mid-management levels. There is no department or domain where operational reporting cannot be used. These reports are the most common type of reports used to describe or measure the existing state.
A simple example of assessing metrics is the expense report of a utility bill that shows energy consumption by day, month, or year. This report is based on the services and electricity already consumed by the consumer. These include multi-dimensional analysis (OLAP) and OLTP (online transactional processing)/web.
Predictive Reports
The second set of reports are those that use historical data but additionally use a statistics and predictive algorithm to provide a prediction for the near future. Predictive models/algorithms may or may not be available for the subject area as per the pre-defined required confidence interval required.
Predictive reporting is the type of reporting to analyze short-term or long-term business trends so as to discover trends and patterns for the future. This type of reporting uses algorithms to find the short-term probability of an event in the near- or long-term future. Some of the examples are as follows:
• Navigate through major trade supply-chain disruption or sudden spikes in demand.
• In Fast Moving Consumer Goods (FMCG), Predict the flow of materials across supply-chain to improve the efficiency of inventory and operations.
• In FMCG, predict demand of FMCG final product. In traditional clinical processes, use trial simulation without recruiting patients.
• In health care, predict a patient’s length of stay and help hospitals save money per patient every year. Some additional areas of benefit are disease identification and diagnosis, medical imaging, and drug discovering and manufacturing.
Some of the algorithms are neural networks, logistic regressions, linear regressions, naïve Bayer classifiers, k-near neighbors’ algorithms, random forest, k-means clustering, regression analysis, decision tree learning, gradient-boosted models, boosting, and more. Even though there are multiple algorithms, the application of these algorithms changes use case to use case.