502
L. Pimentel da Silva and F. T. de Souza
with less than 20,000 inhabitants. The municipality with the least arborization was
Campina do Simão.
The total morbidity of respiratory diseases (RD) per 1000 inhabitants in the state
is higher than that of circulatory diseases (CD). The maximum number of morbidities
of RD per 1000 inhabitants is nearly double the morbidities of CD. Although PCV for
both RD and CD are high, the standard deviation is not so high, and the histograms,
Fig. 2d, e, shows a better distribution of values than for DD and ARB. For RD,
nearly 250 municipalities present up to 20 cases of morbidity per 1000 inhabitants.
Meanwhile, for CD, more than 300 municipalities present up to 15 cases.
Linear correlation Linear correlation was the first multivariate analysis performed.
Before processing, the multivariate analysis variables were standardized. This
avoided the influence of the magnitude and different measured units among variables (Vicini 2005). The whole matrix consists of 9 variables (as described in Box 1)
and 399 instances (number of municipalities in the State of Paraná). There were
no missing values. A linear coefficient near +1 or –1 represents a perfect positive
or negative linear correlation, respectively. Table 2 presents the matrix of linear
coefficients of correlation between indicators. The diagonal is always 1 (one) and
shows the linear coefficient of correlation between the same variables. VHCLS and
M_HDI showed the highest linear coefficient of correlation, followed by the one
between RD and CD; in both cases it was positive. The higher the municipal human
development index, the more vehicles; and the higher the number of morbidities of
respiratory diseases, the higher the number of morbidities of circulatory diseases.
The linear coefficient of correlation was negative between M_HDI and morbidities
of respiratory and circulatory diseases.
The green infrastructure indicator (ARB) presented a negative linear correlation
with DD: the more inhabitants per square kilometer, the lower the number of urban
households in streets with trees. Meanwhile, it was positive between green infrastructure and morbidities of respiratory diseases: the greater the arborization, the higher
the number of morbidities of respiratory diseases. Both RD and CD presented a
negative linear correlation with sanitation and urbanization: the higher the sanitation
and urbanization, the lower the number of morbidities of respiratory and circulatory
diseases. It was unexpected that, although with a low value, VHCLS presented a
negative correlation with the morbidities of respiratory diseases.
Cluster Analysis Figure 3 shows the dendrogram resulted from the hierarchal tree
clustering analysis, built considering the single linkage of the amalgamation rule
and Euclidean distance as the distance measure. The Y-axis and the vertical lines
represent the unified groups by descending order of similarity. The lines along the
X-axis indicate the distance between the formed groups. The dendrogram should
be interpreted from the top to the bottom. Two distinct “branches” are observed in
Fig. 3: Group 1 (G1), formed by DD, GDP, M_HDI, VHCLS, URB, ARB, SNT, and
Small II. Between 20,001 and 50,000 inhabitants.
Medium. Between 50,001 and 100,000 inhabitants.
Large. Between 100,001 and 900,000 inhabitants.
Metropolis. Above 900,000 inhabitants.
L. Pimentel da Silva and F. T. de Souza
with less than 20,000 inhabitants. The municipality with the least arborization was
Campina do Simão.
The total morbidity of respiratory diseases (RD) per 1000 inhabitants in the state
is higher than that of circulatory diseases (CD). The maximum number of morbidities
of RD per 1000 inhabitants is nearly double the morbidities of CD. Although PCV for
both RD and CD are high, the standard deviation is not so high, and the histograms,
Fig. 2d, e, shows a better distribution of values than for DD and ARB. For RD,
nearly 250 municipalities present up to 20 cases of morbidity per 1000 inhabitants.
Meanwhile, for CD, more than 300 municipalities present up to 15 cases.
Linear correlation Linear correlation was the first multivariate analysis performed.
Before processing, the multivariate analysis variables were standardized. This
avoided the influence of the magnitude and different measured units among variables (Vicini 2005). The whole matrix consists of 9 variables (as described in Box 1)
and 399 instances (number of municipalities in the State of Paraná). There were
no missing values. A linear coefficient near +1 or –1 represents a perfect positive
or negative linear correlation, respectively. Table 2 presents the matrix of linear
coefficients of correlation between indicators. The diagonal is always 1 (one) and
shows the linear coefficient of correlation between the same variables. VHCLS and
M_HDI showed the highest linear coefficient of correlation, followed by the one
between RD and CD; in both cases it was positive. The higher the municipal human
development index, the more vehicles; and the higher the number of morbidities of
respiratory diseases, the higher the number of morbidities of circulatory diseases.
The linear coefficient of correlation was negative between M_HDI and morbidities
of respiratory and circulatory diseases.
The green infrastructure indicator (ARB) presented a negative linear correlation
with DD: the more inhabitants per square kilometer, the lower the number of urban
households in streets with trees. Meanwhile, it was positive between green infrastructure and morbidities of respiratory diseases: the greater the arborization, the higher
the number of morbidities of respiratory diseases. Both RD and CD presented a
negative linear correlation with sanitation and urbanization: the higher the sanitation
and urbanization, the lower the number of morbidities of respiratory and circulatory
diseases. It was unexpected that, although with a low value, VHCLS presented a
negative correlation with the morbidities of respiratory diseases.
Cluster Analysis Figure 3 shows the dendrogram resulted from the hierarchal tree
clustering analysis, built considering the single linkage of the amalgamation rule
and Euclidean distance as the distance measure. The Y-axis and the vertical lines
represent the unified groups by descending order of similarity. The lines along the
X-axis indicate the distance between the formed groups. The dendrogram should
be interpreted from the top to the bottom. Two distinct “branches” are observed in
Fig. 3: Group 1 (G1), formed by DD, GDP, M_HDI, VHCLS, URB, ARB, SNT, and
Small II. Between 20,001 and 50,000 inhabitants.
Medium. Between 50,001 and 100,000 inhabitants.
Large. Between 100,001 and 900,000 inhabitants.
Metropolis. Above 900,000 inhabitants.