Savitsky-Golay and Moving Average smoothing in SignalProcessing.
Data Analysis Library
Support for LINQ queries on data frames, vectors and matrices.
New groupings for aggregations include: fixed and expanding windows,
partitions, groupings on value and quantiles, 2D pivot tables,
and resampling.
A large set of aggregators and aggregator functions has been added
with efficient implementations for specific types of groupings
(e.g. moving averages).
New generic Descriptives class for collecting descriptive
statistics of vectors.
Indexes on ordered types now support lookup nearest. Likewise,
data frames with such indexes now support join on nearest.
New Recurrence type lets you specify date/time patterns
for use in, for example, resampling of data frames.
Vector and Matrix Library
All vector and matrix classes are now generic, including sparse matrices and complex versions.
New static Vector and Matrix classes remove the need to specify the element type
as a generic type argument.
New mutability options now let you create read-only vectors, and writable vectors with
copy-on-write semantics.
Most operations on vectors and matrices are available in three
forms: as a static method that returns a new value, as a static
method that returns the value in a supplied object, and as
an instance method that modifies the object in-place.
Operator methods are named consistently for static calls,
in-place and out-of-place updates.
Relational operators are now supported for generic matrices.
The native libraries have been upgraded to Intel MKL version 11.3
Update 2.
The native libraries now support Conditional Numerical
Reproducibility.
The CUDA libraries for 64 bit have been upgraded to CUDA
version 7.5.
A fully managed implementation of the linear algebra library
for single-precision floating-point numbers was added.
Statistics Library
Descriptions tailored to working in an interactive environment
for statistical models, hypothesis tests and other objects.
Statistical models have been integrated with the DataFrame library.
The variable types (NumericalVariable, CategoricalVariable,
DateTimeVariable) have been replaced with generic vectors.
Categorical variables are expanded into indicator variables
when necessary. Several encoding schemes are available.
Models can be persisted in a form suitable for deployment in
predictive modeling applications.
Models can be specified using R-style formulas.
Regression models are now robust against collinearity.
New distribution classes have been added: truncated distributions,
inverse Weibull.