Three generalizations of the synthetic control method (which has
already an implementation in package 'Synth') are implemented: first,
'MSCMT' allows for using multiple outcome variables, second, time series
can be supplied as economic predictors, and third, a well-defined
cross-validation approach can be used.
Much effort has been taken to make the implementation as stable as possible
(including edge cases) without losing computational efficiency.
A detailed description of the main algorithms is given in
Becker and Klößner (2018)
Changes in version 1.3.1 (2018-01-04)
some fixes/increased robustness for edge cases (e.g. single predictor)
progress bar fixed (thereby eliminating compiler warnings)
Changes in version 1.3.0 (2017-08-17)
point size is now settable in ggplot.mscmt.
lp's in ill-conditioned problems are now tackled with the help of various lp solvers to improve robustness. Packages Rglpk and lpSolveAPI are required now.
'no sunny donors' case fixed.
some small bug fixes in ggplot.mscmt.
Changes in version 1.2.0 (2017-01-30)
reasonable default parameters and documentation for outer optimizers have been added.
improved cluster support: (malfunctioning) load balancing with parLapplyLB replaced by load balancing via clusterApplyLB.
Changes in version 1.1.0 (2016-11-29)
newly supported outer optimizer "none" to implant 'artificial' solutions (fixed w and optionally fixed v).
many new outer optimizers are now supported, more documentation will follow.
function compare() for comparison of estimation results, including corresponding updates for print.mscmt and ggplot.mscmt.
names of agg.fns (component of mscmt results) are now set automatically.
user-defined agg.fns are now automatically exported to the cluster.
changed default optimizer to DEoptC.
predictor weights are not printed any more as verbose output in estimations.
some packages were moved from "Imports" to "Suggests".
number of calls to inner optimizer is now counted for benchmarking purposes.
package startup message added.
objective function weights alpha/beta/gamma are now treated correctly.
calls to the lp solver are now more robust.
some typos in vignettes have been corrected.
all data had to be sorted chronologically, this is now optional.
improveSynth reported wrong 'True' W*(V) if W*(V) was not optimal.
Changes in version 1.0.0 (2016-07-25)