PAF Version 2.6.1 Release Notes
The Predictive Analytics Framework v 2.6.1 features include:
- A complete Linux build environment (gcc, make, binutils)
- Our increased security features (iptables, fail2ban, custom configurations)
- OptimR Module for R version 3.6.0 (Planting of a Tree)
- 99% of all CRAN installed. (13530 Packages)
- Open Source (free) RStudio Shiny Server 184.108.40.2063 on port 3838
- RStudio Desktop version 1.2.1335
- RStudio Server (free) version 1.2.1335 on port 8787
(user ubuntu has a default password the AWS instanceID)
- Microsoft JDBC Driver 6.0 for SQL Server in the directory /usr/local/drivers/
- Licensed Intel Distribution for Python (versions 2.7.14 & 3.6.3)Performance accelerations: Scikit-learn with Intel® Data Analytics Acceleration Library (Intel® DAAL), fast Fourier transforms in SciPy and Numpy, universal functions (ufuncs) can use multiple cores and Single Instructions Multiple Data (SIMD), and neural network enhancements for pyDAAL
Tech preview: High-level Python API for Intel® DAAL
OptimR Module for R version 3.6.0 (Planting of a Tree)
PAF version 2.6.1 includes the latest release of R (3.6.0) optimized and compiled with open source Message Passing Interface (Open MPI 4.0) support. Our code examples run twice as fast as the same code running under conventionally compiled R on only a single core. The performance is likely even better in multi-core environments. This could have a considerable impact on your calculation, but, like any computing, it depends on exactly what you are doing.
As with any interpreted language invoked under Linux the way to use the compiled R we provide is through use of the shebang syntax.
To use PAF optimized R this start scripts with this shebang line:
Note: the version of R we have added to the operating system is under /usr/local/.
Intel Corporation Python 2.7.14 can be found at:
Intel Corporation Python 3.6.3 can be found at: