Geoscience Reference
In-Depth Information
Table 8.3 Interval classi
cations by joint probabilities
Balanced Interval
0.5 Reductions
Interval
Max and Min
Interval
Lower
Upper
Lower
Upper
Lower
Upper
Car1
3
4
3
4
1
5
Car2
3
4
2
4
1
5
Car3
4
4
4
5
2
5
Car4
4
4
4
5
2
5
Car5
3
3
3
3
1
5
Car6
3
3
2
3
1
5
Car7
4
4
4
4
2
5
Car8
3
3
3
4
1
5
Car9
4
4
3
4
2
4
Car10
4
4
4
4
2
5
Car11
3
4
2
4
1
5
Car12
4
4
3
4
2
5
Car13
3
3
3
4
2
5
Car14
4
4
3
4
2
5
Car15
3
4
3
4
1
5
Car16
4
4
3
4
1
5
Car17
4
4
3
4
1
5
Car18
3
3
3
4
1
5
Car19
4
4
3
4
1
5
Car20
3
3
3
4
1
5
It can be seen in Table
8.3
, as the range of the intervals increases, that the
classi
cation of some cars can move to a better or a worse class. For instance,
looking at the central columns, it can be detected a possibility of Car8, Car13,
Car18 and Car20 to improve their classi
cations from C
3
to C
4
and a possibility of
Car9, Car16, Car17 and Car19 to suffer a downgrade from C
4
to C
3
.
8.6 Classification of Countries by HDI Criteria
An example of generation of pro
ed
may be given by revisiting the analysis of human development of countries on the
basis of the four criteria: longevity, mean years of schooling, expected years at
school and per capita income. Suppose that, instead of determining a distance to a
global frontier, we prefer to classify the countries in a small number of classes.
Fixing in
les by the sample of alternatives to be classi
five the number the classes, one pro
le for each class may be derived from
the deciles of the sample of observed values.
By similar procedures, other two pro
les may be derived for each class. In
addition to the vectors of coordinates given by approximate values of the
first, third,
Search WWH ::
Custom Search