README file for ks


New v1.5

1. New functions to compute pilot functinal estimators hat{psi}_r(g).
These are exact, and are more efficient than binned estimators for small
samples (~100), and are available in d > 4.

2. Added plot3d() capabilities from rgl to 3-d plot - removing own axes
drawing functions.

3. Made changes to specifying line colour (col rather than lcol) in plot.kde, 
plot.kda.kde and partition class colour (partcol) in plot.kda.kde. 

4. Added unconstrained pilot selectors for d = 1, ..., 5 as option `pilot="unconstr"' in `Hpi'.

5. Added plug-in selectors for 1-d data (exactly the code for dpik from KernSmooth). For KDE, this is `hpi', for KDA, this is 'hkda(..., bw="plugin")'.
 
6. Added distribution functions for 1-d KDEs: dkde, pdke, qkde, rkde.

7. Dependence on Matrix package removed. Added dependence on combinat.

8. Added normal mixture (A)MISE-optimal selectors: hamise.mixt, hmise.mixt, Hamsie.mixt, Hmise.mxt.

9. >=1.5.10. Added drvkde (kernel density derivative estimator 1-d) from feature using Matt Wand's code.

10. 1.5.11. Fixed small bugs in drvkde, vech, Hpi(, pilot="unconstr").  
 
---------------------------------------------------------------------------
New in v1.4

1. Hscv.diag() and Hbcv.diag() added for completeness.

2. compare.kda.cv() and compare.cv() modified to improve speed.

3. Filled contour plots added as disp=filled option in plot.kde().

4. Binned estimation implemented for KDE with diagonal selectors and pilot functional
estimation with diagonal selectors.

5. Vignette illustrating 2-d KDE added.


---------------------------------------------------------------------------
New in v1.3

1. Multivariate (for 3 to 6 dimensions inclusive) bandwidth selectors added
for `Hpi', `Hpi.diag' (plugin), `Hlscv', `Hlscv.diag' (least squares cross
validation) and `Hscv' (smoothed cross-validation).  So this means that
these selectors can also be used in kernel discriminant analysis. (NB:
because `Hbcv' and `Hbcv.diag' (biased cross validation) performed poorly
for 2-dimensions, these have not been implemented in the higher dimensions).

2. RGL-type plots added for 3-d data. See `plot.kde' for kernel density estimates
and `plot.dade' for discriminant analysis (both kernel and parametric).
Note. in versions <=1.3.3, the specification of contours for 3-d was the reverse
to that for 2-d. For v >=1.3.4, there are now the same. 3-d axes added.

3. Fixed small bug in compare.kda.cv() and compare.kda.diag.cv().


---------------------------------------------------------------------------
New in v1.2

1. The child mortality data set `unicef' is used in the examples now. See `unicef'
for more details.

2. Package has been checked by CRAN testers and has been accepted on the CRAN
website. To pass all the necessary checks involved some internal programming
changes but has not affected the user interface.


---------------------------------------------------------------------------
New in v1.1

1. The `adapt' library is no longer required. This was formerly used on
the functions for integrated squared error computations (i.e. `ise.mixt'
and `iset.mixt').

2. Kernel discriminant analysers are now available. See help on
`kda' and `Hkda'. Parametric (i.e. linear and quadratic) discriminant analysers
are accessed using  `pda'.

3. Cross validation bandwidth selectors are now available. They are `Hbcv' and
`Hbcv.diag' for biased cross validation, `Hlscv' and `Hlscv.diag" for least squares
cross validation and `Hscv' for smoothed cross-validation.

4. S3 type objects have been introduced.  The output from `kde' are `kde' objects.
The output from `kda.kde' and `pda.pde' are `dade' objects. The major effect of this
is that the corresponding plot functions (i.e.`plot.kde' and `plot.dade') will
be called automatically by invoking just `plot'.
