Information Technology Reference
In-Depth Information
Figure 20. Probability density for total charges concerning immune disorder and septicemia
if (septicemia=
0
and immune=
0
and pneumonia=
0
) then diagnosis='none';
if (septicemia=
0
and immune=
1
and pneumonia=
0
) then diagnosis='immune';
if (septicemia=
1
and immune=
0
and pneumonia=
0
) then diagnosis='septicemia';
if (septicemia=
0
and immune=
0
and pneumonia=
1
) then diagnosis='pneumonia';
if (septicemia=
1
and immune=
1
and pneumonia=
0
) then diagnosis='sepimm';
if (septicemia=
1
and immune=
0
and pneumonia=
1
) then diagnosis='seppne';
if (septicemia=
0
and immune=
1
and pneumonia=
1
) then diagnosis='immpne';
if (septicemia=
1
and immune=
1
and pneumonia=
1
) then diagnosis='all';
run;
proc sort
data=work.immseptpnecombine;
by diagnosis;
run;
proc kde
data=work.immseptpnecombine;
univar los/gridl=
0
gridu=
20
method=srot bwm=
2.5
out=nis.kdelosimmseptpne;
univar totchg/gridl=
0
gridu=
50000
method=srot bwm=
.8
out=nis.kdechgimmseptpne;
by diagnosis;
run;
The result again shows that the condition of septicemia dominates the other two conditions (Figures
21 and 22). The immune disorder with pneumonia is shown in the two figures to utilize more healthcare
resources than either condition alone.
We will use these same three conditions in Chapter 3 to show how statistics, including data sum-
maries and regression models can be used to examine the data; these analyses will be in contrast to the
visualization techniques discussed in this chapter.
Search WWH ::
Custom Search