v0.2.0

v0.2.0

Breaking changes(!):

• Renamed exported functions: ctsamplen, ctpower, besamplen, bepower, ets (now in lower case).
• nca() -> nca!() - renamed;
• Changes in ctsamplen() in one group estimation. In version 0.1.x test mean/proportion denoted as parameter/keyword "a" was a "Reference".
  This situation was inconsistent with two group analysis where "a" - T, "b" - R. Now everywhere "a" means "Test group", "b" means "Reference group".
  Hypothesis testing for difference is A-B everywhere.
• Keyword "logdiff" removed in ctsamplen(), because this was inconsistent with bepower/besamplen, new parameter logscale using instead.
• Keyword "out" removed everywhere. Output via Base.show(io::IO, obj::TaskResult);
• twoProp(), oneProp(), oneMean. twoMeans() removed;
• No submodule politics.

Major changes:

• using DataFrames 19.0+
• Task types for many functions (sample size, power functions, random ets.);
• Sample size and power functions now return TaskResult{T <: CTask};
• descriptive() rewrited;
• DataSet type;
• make DataFrame from DataSet;
• Descriptive type;
• CMH CI split;
• randomtable() return DF
• Include CI in ClinicalTrialUtilities & StatsBase methods for AbstractParameter types.

Other changes:

• Changes in error messages. Now throw(ArgumentError(msg));
• LimitRule for PK;
• RMIN/RMAX/TH/BL for PD;
• Steady-state PK and other PK parameters;
• Descriptive stats add SES and SEK;
• Descriptive performance;
• Test procedures restructured;
• 4x4 design;
• Documentation;
• Minor changes, bugfix, code cleaning, cosmetics.