Pluggable Computational Engines
Inference for QbD employs the R statistical computational and graphic visualization engine as its data-analysis service. R, developed as a global open source project led by leading researchers in applied computation, is deemed the premier software environment for R&D in data analysis, statistics and modeling.
Inference Data-Analysis Services uses a pluggable architecture that enables the user to select an R engine deployment that meets their specific business needs. Deployment choices include the following:
- Desktop-configured R-installation
- Centralized server R-installation to ensure consistency and compliance in data-analysis across the entire QbD team
- Multiple distributed server R-installations (server farm/cluster) to permit distributing large data analysis loads in the enterprise
- Multiple parallel processors (multiprocessor) R-installation to permit a dramatic decrease in time to completion of large QbD computational tasks
Most customers deploy the open-source R-Project (http://www.R-project.org) installation of R on their desktop as their data-analysis service. Those customers that need their R-installation validated under FDA guidelines may chose to use one of several alternative commercial R-installations, which include support for software validation. Examples of such alternatives include RStat (Random Technologies, http://www.random-technologies.com), RPro (Revolution Computing, http://revolution-computing.com), and R+ (XLSolutions Corporation, http://www.xlsolutions-corp.com).
Inference Data-Analysis Services may also be deployed using an alternative data analysis platform written in other scripting languages(e.g., Python, MATLAB, SAS, .NET). Thus, “best practices” software from a variety of scripting languages and data analysis platforms can be leveraged and reused.