<s>
Robust	B-General_Concept
decision-making	I-General_Concept
(	O
RDM	O
)	O
is	O
an	O
iterative	O
decision	O
analytics	O
framework	O
that	O
aims	O
to	O
help	O
identify	O
potential	O
robust	O
strategies	O
,	O
characterize	O
the	O
vulnerabilities	O
of	O
such	O
strategies	O
,	O
and	O
evaluate	O
the	O
tradeoffs	O
among	O
them	O
.	O
</s>
<s>
One	O
source	O
of	O
the	O
name	O
"	O
robust	B-General_Concept
decision	I-General_Concept
"	O
was	O
the	O
field	O
of	O
robust	O
design	O
popularized	O
primarily	O
by	O
Genichi	O
Taguchi	O
in	O
the	O
1980s	O
and	O
early	O
1990s	O
.	O
</s>
<s>
Jonathan	O
Rosenhead	O
and	O
colleagues	O
were	O
among	O
the	O
first	O
to	O
lay	O
out	O
a	O
systematic	O
decision	O
framework	O
for	O
robust	B-General_Concept
decisions	I-General_Concept
,	O
in	O
their	O
1989	O
book	O
Rational	O
Analysis	O
for	O
a	O
Problematic	O
World	O
.	O
</s>
<s>
Robust	B-General_Concept
decision-making	I-General_Concept
(	O
RDM	O
)	O
is	O
a	O
particular	O
set	O
of	O
methods	O
and	O
tools	O
developed	O
over	O
the	O
last	O
decade	O
,	O
primarily	O
by	O
researchers	O
associated	O
with	O
the	O
RAND	O
Corporation	O
,	O
designed	O
to	O
support	O
decision-making	O
and	O
policy	O
analysis	O
under	O
conditions	O
of	O
deep	O
uncertainty	O
.	O
</s>
<s>
While	O
often	O
used	O
by	O
researchers	O
to	O
evaluate	O
alternative	O
options	O
,	O
RDM	O
is	O
designed	O
and	O
is	O
often	O
employed	O
as	O
a	O
method	O
for	O
decision	B-Application
support	I-Application
,	O
with	O
a	O
particular	O
focus	O
on	O
helping	O
decision-makers	O
identify	O
and	O
design	O
new	O
decision	O
options	O
that	O
may	O
be	O
more	O
robust	O
than	O
those	O
they	O
had	O
originally	O
considered	O
.	O
</s>
<s>
This	O
ordering	O
provides	O
cognitive	O
benefits	O
in	O
decision	B-Application
support	I-Application
applications	O
,	O
allowing	O
stakeholders	O
to	O
understand	O
the	O
key	O
assumptions	O
underlying	O
alternative	O
options	O
before	O
committing	O
themselves	O
to	O
believing	O
those	O
assumptions	O
.	O
</s>
<s>
Robust	B-General_Concept
decision	I-General_Concept
methods	O
seem	O
most	O
appropriate	O
under	O
three	O
conditions	O
:	O
when	O
the	O
uncertainty	O
is	O
deep	O
as	O
opposed	O
to	O
well	O
characterized	O
,	O
when	O
there	O
is	O
a	O
rich	O
set	O
of	O
decision	O
options	O
,	O
and	O
the	O
decision	O
challenge	O
is	O
sufficiently	O
complex	O
that	O
decision-makers	O
need	O
simulation	O
models	O
to	O
trace	O
the	O
potential	O
consequences	O
of	O
their	O
actions	O
over	O
many	O
plausible	O
scenarios	O
.	O
</s>
<s>
RAND	O
Corporation	O
has	O
developed	O
CARS	O
for	O
exploratory	O
modeling	O
and	O
the	O
sdtoolkit	O
R	B-Language
package	O
for	O
scenario	O
discovery	O
.	O
</s>
<s>
The	O
EMA	O
Workbench	O
,	O
developed	O
at	O
Delft	O
University	O
of	O
Technology	O
,	O
provides	O
extensive	O
exploratory	O
modeling	O
and	O
scenario	O
discovery	O
capabilities	O
in	O
Python	B-Language
.	O
</s>
<s>
OpenMORDM	O
is	O
an	O
open	O
source	O
R	B-Language
package	O
for	O
RDM	O
that	O
includes	O
support	O
for	O
defining	O
more	O
than	O
one	O
performance	O
objective	O
.	O
</s>
<s>
Rhodium	O
is	O
an	O
open	O
source	O
Python	B-Language
package	O
that	O
supports	O
similar	O
functionality	O
to	O
the	O
EMA	O
Workbench	O
and	O
to	O
OpenMORDM	O
,	O
but	O
also	O
allows	O
its	O
application	O
on	O
models	O
written	O
in	O
C	O
,	O
C++	O
,	O
Fortran	O
,	O
R	B-Language
and	O
Excel	O
,	O
as	O
well	O
as	O
the	O
use	O
of	O
several	O
multi-objective	O
evolutionary	O
algorithms	O
.	O
</s>
