Version numbering:

0 for alpha (status); 
1 for beta (status); 
2 for release candidate; 
3 for (final) release

bda_1.2.7-31:  algorithms for binned data analysis,
  data with rounding errors.

bda_2.0.11-4: 
  (1) to update binning function and weighting function.  
  (2) to add lprde (no SCB included yet).

2012/11/05:  
  (1) to add survival function estimation, 
  (2) to provide a function to weight data for users. 
  (3) to update keywords in the Rd files.
  (4) to add bandwidth selector 'erd'
  (5) add an bandwidth option for 'aMAE' or 'MAE'.

2012/11/10:
  (1) fixed a bug of different length in .fftkde
  (2) fixed a bug in computing WIQR
  (3) add to binning an option such that if binwidth is given
      but 'lb' is missing, we set lb = min(x)-binwidth/2.

2012/11/14:
  (1) remove startup message
  (2) rewrite qiwr function
  (3) fix a bug in wkde --> "bw"/"bandwidth"
  (4) remove "amae" from .wkde

2012/11/19:
  (1) add range to "mae" and make sure the lower limit is non-negative

2013/02/08:
  (1) add 'bdde' function to estimate from data that are prebinned --
	divided in to several groups (intervals).
  (2) add two more distribution families other than "weibull" -- 
	ewd and datum. 
2013/02/14: add function ROC
2014/03/19: add plot option to npr(smooth)
2014/05/06: rewrite/develop algorithms for density estimation based 
	on binned data (histo-smoothing). 

2014/05/20: 3.2.0-4 (1) fixed a bug in bootKDE when iter>limit.

2014/05/20: 3.2.0-5 (1) remove CUSUM, ROC, PermTest, and source 
	C/Fortran codes. (2) MEDE (redistributing weights for density 
	estimation under measurement errors.) (3) add dependent package
	MASS for ginv.

2014/05/20: 3.2.0-6 (1) remove MEDE. (2) remove dependency on MASS.

2014/06/10: 4.0.1 (consider difference cases of grouped data: 
	top-code/non-top-coded, equal bin width/unequal bin width; 
	bounded data/unbounded.
	
2014/06/16: add FMKL GLD fitting.  Weibull fitting.

2014/06/18: add gof.test, add fit.GB, 
        add print/plot/lines methods for FMKL.

2015/03/01: check gof for GLD/GB for binned data, add an optimaize 
	estimate based on GOF test.

2015/04/03: move some of the algorithms developed from bstats.

2015/07/29: add histospline, smkde, and ofc data 

2015/07/30: Rd file updated for the Dagum distribution.

2017/04/03: add a npr function for deep sequencing data analysis. errors-in-variables and simultaneuous confidence bands.

2017/09/12: change outputs of .rescale.mm and normalize.NGS.MM.

2017/09/13: rewrite normalize.NGS.MM(...).  Using QN-like method -- two stage normalization.

2017/09/29: copula fitting
2017/10/03: add an internal function to choose initial values to fit binned data using normal mixture.
2017/10/09: update fit.NGS function; update npreg function.
2017/10/13: rewrite NGS data binning function bin.NGS and fit.NGS function
2017/10/14: add back mixed=TRUE option to fit.NGS.  If sLog=0 and mixed=FALSE, 
            fit top-coded-type histogram.
2017/10/15: fitting lognormal using MME, MLE based adjusted data, and MLE based
	    on binned data.
2017/10/16: add option 'binned' to tell whether MLE was refined based on 
	    binned data.
2017/10/25: best fit by change number of components (k).
2017/10/26: add k=1 in fit.NGS and change the summary function accordingly.
            add heterogeneity test in fit.NGS
2017/10/27: add to fit.lognormal a test H0: mean = 0 vs Ha: mean > 0.
2017/10/28: revised fit.lognormal (check parameters in nlm() function
2017/11/14: add a test to deg.NGS showing whether the data is under-detection 
            level or not (using binomial distribution to compute the exact 
            probabilities.
2017/11/15: revise lognormal fitting function.
