Companies can choose from a variety of software tools to perform advanced analytics.
In principle, there are four distinct classes of software:
Use-case-specific solutions that offer easy access to sometimes complex statistical and mathematical analysis methods;
Business intelligence solutions extended by advanced analytics functions;
Advanced analytics platforms and traditional data mining software;
Proprietary developments (e.g., using the extensive and license-free “R libraries”)
In addition to the algorithms and procedures required for the application, use-case-specific solutions also offer business user-friendly interfaces. These conceal complicated models and their corresponding calculations from the user.
Using pre-defined logic in the form of task-specific modules, software providers are making their systems as user-friendly as possible. For example, business departments needing to carry out simple, regular analyses such as shopping cart evaluations, customer classifications or price optimizations have a range of tools with predefined templates, content and corresponding user interfaces to choose from.
However, when business requirements become more complex (e.g., there are not three, but thirty-three characteristic values to be included in a data mining model), the picture becomes altogether different.
Also, if the model has to be adapted (e.g., new variables are generated), the limitations of use-case-specific solutions become apparent. The models must then be further developed, which is no trivial task, by experts with the necessary skills. Functional and easy-to-use “advanced analytics platforms” support this process of finding the right data mining model.
Products historically based mainly on business intelligence (i.e. visualizing and evaluating data using dashboards, reports or analyses), are increasingly being equipped with advanced analysis functions to visualize not only the current state of affairs, but also to calculate complex statistical models or predictions. The power of these functions varies considerably. Most of these BI products have a form editor in their reporting and analysis tools.
This allows the user to store simple mathematical calculations in reports or analyses. As well as determining the minimum or maximum value within an analysis, the report designer can, for example, also store a calculation which uses exponential smoothing to determine a forecast value such as, for example, sales in the coming months.
However, business intelligence software has limitations in more complex predictive analytics tasks: modeling and implementation of more advanced application-specific tasks, neural networks, decision trees and text mining usually require external software (e.g., advanced analytics platforms).