Urban Management: Learning from Green Infrastructure …
499
Apart from these previous analyses, a first attempt was made to mine rules that
would explain respiratory and circulatory morbidities through the socioeconomic
and environmental variables, including the “ARB”, the percentage of urban households in streets with trees, which was taken as the green infrastructure indicator.
Morbidity levels were computed by considering terciles (low, medium, and high).
The classification was based on the association rules algorithm “CBA”, described in
Liu et al. (1998). In the first step, the algorithm scans the whole data set to find a
set of classification rules with strong support (union of input and output variables).
Then, the algorithm applies this set of rules to predict morbidity and to validate the
rules previously established. The result of the classifier validation is presented in a
confusion matrix where the main diagonal shows correct classifications, and misclassifications are shown outside the diagonal. The correct classification rules measure
the level of accuracy, which is given by the sum of the correctly classified samples
divided by the total number of instances sampled.
3 Results and Discussion: Descriptive Statistics, Linear
Analysis, Cluster Analysis, Factor and Main Component
Analysis, Classification Models
Descriptive Statistics Table 1 shows the descriptive statistics for all variables. Except
for M_HDI, all variables present a large range of variation, albeit rating some of them
per 1000 inhabitants. The standard deviation of DD is greater than its mean, showing
that DD is heterogeneous. The largest value of DD was 4.027,00, for Curitiba (the
main city), while the smallest was equal to 3.31 for the municipality of Alto Paraíso,
in the Northwest part of the State. The histogram (Fig. 2a) demonstrates that nearly
160 municipalities, out of 399, have demographic densities ranging from about 19 to
35 inhabitants per square kilometer. Considering the two most frequent ranges of DD,
about 70% of all municipalities have demographic densities ranging from about 3 to
nearly 35 inhabitants per square kilometer. If a threshold of DD is established as 35,
above this value, municipalities’ demographic density seems to be better distributed.
The largest Pearson’s coefficient of variation (PCV) is for demographic density (DD),
followed by the percentage for sanitation (SNT), which is 84.08%. As expected, the
smallest PCV value is for M_HDI.
Regarding arborization (ARB), which is taken as the green infrastructure indicator, about half of the municipalities fall into the 90–100% class. The maximum
arborization rate, 100%, was found for the municipalities of Arapua, Lobato, Miraselva, Santa Fé and Santa Inês, located in the Mid-North Mesoregion of the State of
Paraná, and for Guaporema, Indianopolis and Nova Aliança do Ivaí, located in the
Northwest Mesoregion of the State. These are all classified as small I
2 municipalities
2 The Brazilian Institute of Geography and Statistics (IBGE) classifies Brazilian municipalities
according to their total population into:
Small I. up to 20,000 inhabitants.
499
Apart from these previous analyses, a first attempt was made to mine rules that
would explain respiratory and circulatory morbidities through the socioeconomic
and environmental variables, including the “ARB”, the percentage of urban households in streets with trees, which was taken as the green infrastructure indicator.
Morbidity levels were computed by considering terciles (low, medium, and high).
The classification was based on the association rules algorithm “CBA”, described in
Liu et al. (1998). In the first step, the algorithm scans the whole data set to find a
set of classification rules with strong support (union of input and output variables).
Then, the algorithm applies this set of rules to predict morbidity and to validate the
rules previously established. The result of the classifier validation is presented in a
confusion matrix where the main diagonal shows correct classifications, and misclassifications are shown outside the diagonal. The correct classification rules measure
the level of accuracy, which is given by the sum of the correctly classified samples
divided by the total number of instances sampled.
3 Results and Discussion: Descriptive Statistics, Linear
Analysis, Cluster Analysis, Factor and Main Component
Analysis, Classification Models
Descriptive Statistics Table 1 shows the descriptive statistics for all variables. Except
for M_HDI, all variables present a large range of variation, albeit rating some of them
per 1000 inhabitants. The standard deviation of DD is greater than its mean, showing
that DD is heterogeneous. The largest value of DD was 4.027,00, for Curitiba (the
main city), while the smallest was equal to 3.31 for the municipality of Alto Paraíso,
in the Northwest part of the State. The histogram (Fig. 2a) demonstrates that nearly
160 municipalities, out of 399, have demographic densities ranging from about 19 to
35 inhabitants per square kilometer. Considering the two most frequent ranges of DD,
about 70% of all municipalities have demographic densities ranging from about 3 to
nearly 35 inhabitants per square kilometer. If a threshold of DD is established as 35,
above this value, municipalities’ demographic density seems to be better distributed.
The largest Pearson’s coefficient of variation (PCV) is for demographic density (DD),
followed by the percentage for sanitation (SNT), which is 84.08%. As expected, the
smallest PCV value is for M_HDI.
Regarding arborization (ARB), which is taken as the green infrastructure indicator, about half of the municipalities fall into the 90–100% class. The maximum
arborization rate, 100%, was found for the municipalities of Arapua, Lobato, Miraselva, Santa Fé and Santa Inês, located in the Mid-North Mesoregion of the State of
Paraná, and for Guaporema, Indianopolis and Nova Aliança do Ivaí, located in the
Northwest Mesoregion of the State. These are all classified as small I
2 municipalities
2 The Brazilian Institute of Geography and Statistics (IBGE) classifies Brazilian municipalities
according to their total population into:
Small I. up to 20,000 inhabitants.