Building Density Level of Urban Slum Area in Jakarta

Currently, the number of urban residents is increasing and some of the urban population live in slum areas. Therefore, identifying the characteristics of slum areas has become crucial. This study aimed to identify more specific slum locations in Jakarta through the pattern of building densities analysis between case studies of neighborhoods association (RT) in 15 hamlets (RW) that classified as heavy slums category. This study also attempted to determine the relation between building density levels in the slum area and Jakarta spatial detail planning. This study engaged the Cluster and Outlier Analysis (Anselin Local Moran's I) method. This study also observed socio-economic factors of citizen census data based the Dasawisma Census of Family Welfare Empowerment in 2019. The result shows that slum locations that had direct neighbors towards areas which was designated at spatial detail planning as industrial zones and ware housing areas as well as office, trade and service zones, obtained a higher level of building density compared to slum locations that secured neighbors to areas that were designated as housing zones. High economic opportunities provided attraction and affected the growth of slum locations. The results also reveal that slum areas were not a concentrated population with low income and/or low education. Applying cluster and outlier modeling of building density levels of urban slums in Jakarta based on RT cluster level could reveal more specific slum locations and could identify factors that influence the differences of building density levels.

questionnaires towards respondents, revealed that the major factors affecting slums were affordability, cultural, kinship and markets adjacency, and then followed by other factors (Badmos et al., 2020). Friesen et al. (2018) also utilized population data of slum area to indicate that slum development was strongly related to demographic development of a country.
Some studies utilizing building maps and spatial methods to identify urban areas based on building density levels. Buildings data could be used to describe community residence or to describe a business district within down town compared to population data (De Bellefon et al., 2020). High density is also one of the physical characteristics that is frequently used to describe slums. Furthermore, in term of physical characteristics, there are also the small roof size and irregular patterns (Kuffer et al., 2016). Spatial statistical method also measures spatial concentration including high-density buildings. Arribas-Bel et al. (2019) also engaged building density measurements based on machine learning algorithms, proposed to obtain the distribution of building groups with significant values that could reach the minimum building density limit.
De Bellefon et al. (2020) compared building density with engaging building data in France.
Hence, the definition of urban area is characterized by high building density.
The results showed the communal space usage not only in public locations that already remained in spatial planning, but also communal space utilization also occupied a huge of space for roads.
In the same location, Wati (2018) revealed that due to improper housing space conditions, the private space engagement spread out and utilized public space.
Slum issues and spatial detail planning are closely interrelated. This is comformable to the New Urban Agenda which is committed to promote the planning and replenishment of city expansion, includes a commitment to recover slums. Slums in Jakarta were classified using RW (Rukun Warga/hamlet) administrative boundaries and there was still limited information regarding slums in the smaller areas or certain slum locations and limited information to determine the effect of spatial detail planning factors on these locations.
The previous studies concerned in identifying the characteristics of slums based on the population census, field study and remote sensing methods. There have been limited studies focused on slum areas interrelated to spatial detail planning in Jakarta using building density. Therefore, this study aimed to identify more specific slum locations in Jakarta through the pattern of building density analysis between case studies neighborhoods association (RT) in 15 hamlets (RW) that classified as heavy slums category. This study committed to investigate the affiliation between building density level in slum area and Jakarta spatial detail planning. The current research also observed socio-economic factors based on population census data reported by the Dasawisma Census of Family Welfare Empowerment in 2019.

Study Area
The study area employed in this study was Jakarta Municipality. Jakarta Municipality was the capital city which was divided into five regions (Kota Administrasi) composed of 42 sub-district (Kecamatan) that consisted of 262 urban villages (Kelurahan) and divided into 2,718 citizen associations (RW). Each RW was divided into 10-20 neighborhoods (RT). Jakarta is the capital and largest city of Indonesia that covers 662,33 square kilometers. and Central Jakarta with one location namely Tanah Tinggi urban village RW 012.
These 15 locations were considered as heavy slums category were taken as samples, since those were assumed to be locations with the highest slum level compared to others. It was necessary to analyze these fifteen (15) locations as priority locations to improve slums and the analysis results could be utilized as references for other categories. The distribution of fifteen RW locations is presented in Figure 1 with heavy slum legend. Data analysis employed dependent and independent variables. The dependent variable was building density level which was one of the physical indicators of slums (Ministry of Public Works and Indonesian Public Housing-PUPR, 2016). A high level of building density that located within metropolitan cities and large cities is > 250 units/Ha, this parameter is engaged for Jakarta. The parameter value was 5 if the building density was in the range of 76% -100% of the indicator, the value 3 was if the building density was in the range of 51% -75% of the indicator, and the value was 1 if the building density was in the range of 25% -50 of the indicator, a value of 5 was categorized a high score, a value of 3 was categorized a medium score and a value of 1 is categorized a low score.
Tenty Melvianti Legarias et al / GEOSI Vol 5 No 2 (2020) 268-287 The independent variables employed for this study were taken from two sources. The first source was acquired based geographical factors, namely the spatial detail planning map and the second source was socio-economic factor,which was according to Dasawisma data composed of education, level of family activity in the settlement, and income which was represented by the head of the family.

Analysis Method
The research methods used were descriptive analysis and spatial analysis to accomplish the research objectives that related to spatial/regional aspects. Overlay analysis is an operation of GIS to superimpose several layers of a dataset which representing different themes together to analyze or identify the relationships of each layer. Overlay analysis represents a composite map with a combination of various attributes and the geometry of a data set or entity. An overlay is an operation of comparing variables among several scopes. During analysis process with applying overlay method, a new map could be generated which wasa combination two or more layers of input maps results. This current study, overlay analysis was performed between building maps, administrative boundary maps, and the spatial detail planning map.
In addition, Local Moran's I index (LMI) was applied to detect the clusters and outliers based on dependent variables building density level in neighborhoods (RT) administrative boundary maps. This spatial analysis namely Cluster and Outlier Analysis. The analysis was needed to be completed in order to identify the geographical distribution of slum phenomenon and to determine the factors statistically whether result in dependence phenomenon on other regions or it was an independent phenomenon. In order to formulate statistical calculations into clusters and outliers, LMI calculated the Moran I value, z-score, p-value, co-type. Z-score is the standard deviation, and the p-value is the opportunity value of mistrust that spatial patterns are random.
The Local Moran's I statistic of spatial association (Anselin, 1995) is presented below: Where is attribute for feature i, is the mean of the corresponding attribute, , is the spatial weight between feature i and j, and: With n equating to the total numbers of features.
The score for the statistics are computed as: where: The result I value reveals the positive or negative value of the feature. If the value of I was positive, it meant that the feature was part of a cluster with neighboring features that were equally high or low value, whereas if the value of I was negative then the feature was an outlier where the feature had a different value than its neighbors. In both conditions, the p-value must be of very small value, thus it could be considered a significant value. In the sense of this statistic, a maximum p-value of 0.05, could be categorized as significant was set at a 95% confidence level.
Co-type on statistical results was an attribute that distinguished whether a high-value significant cluster (HH), or a low-value significant cluster (LL), and for the outlier category would distinguish whether a feature was surrounded by high-value features (LH) or whether a feature was surrounded by low value features (HL). The Co-type statistical results of RT area conducted in this study is presented on Figure 2.

Distribution of 15 RW's Building Density Level of Heavy Slums
The initial stage was observing the distribution of the building density level of 15 RW's in heavy slum which limited the scope of this study to identify the characteristics of each RW.
Spatial overlay analysis was performed between the administrative boundary map of 15 RW and the building map to calculate the level of building density. The results can be seen on Table 1.      information maps, these locations were worked for commercial zones and burial grounds. Cluster and outlier analysis results map and spatial detail planning map for Kapuk RW 12 can be seen on Figure 5. The Cluster and outlier analysis results could be seen on Table 4.  The area of HH cluster was an average of 81% in the residential zone, all RT's outside this cluster were located within commercial zones and industrial zones. The LH cluster and LL cluster were put in locations that designated as industrial and trade zones, despite people settled in the area. Cluster and outlier analysis results map and spatial detail planning map for Penjaringan RW 008 is presented on Figure 6. The map reveals the impact of industrial zones and commercial zones towards slum areas. The Cluster and outlier analysis results can be seen on Table 5.  there was no area had spatial detail planning for residential zones. The HH cluster was100% in the industrial zone.Cluster and outlier analysis results map and spatial detail planning map for Kalibaru RW 007is presented on Figure 7. This map reveals the impact of industrial upon slum areas. The Cluster and outlier analysis results is shown on Table 6.   Figure 8. This map illustrates the impact of industrial towards slum areas. The Cluster and outlier analysis resultscould be observed on Table 7.   The current study discussed the analysis results of the 6 RW's heavy slum category with two RWs located exactly alongside the coastal. The analysis was performed to reveal the fact that slums did not cover the entire RW area by analyzing patterns of distribution of building densities to smaller areas, that were areas based on RT administrative boundaries.
The results show that slum locations had direct neighbors to areas were designated as industrial zones and warehousing areas as well as office, trade and service zones, obtained a higher level of building density compared to slum locations that had neighbors to areas that were designated as housing zones according to Jakarta spatial detail planning. These results are in accordance with (Roy & Lees, 2020) which revealed attractive economic opportunities would likely attract residents of slums. In addition, (Takyi et al., 2020) stated where all slum areas in their analysis were located around the Central Business District. The result is also in line with (Badmos et al., 2020) that discovered a factor influenced the choice of residence for slum was markets adjacency. Zain et al., (2018) in his research indicated that engaging space for the development of the trade and services area had an impact on the growth of slums.
The expansion of an area or space interm of developing trade and services is proportional to the growth of slums which could also escalate regional microeconomic growth, where slum dwellers operate small-scale businesses (Zain et al., 2018). Slums are growing near strategic areas including business centers, trade, markets, or industries (Badmos et al., 2020;Prianto & Amalia, 2019;Zain et al., 2018). Slums also contribute to the development of a nearby business district, since this sector could provide man power to support the operations of the business district (Ray, 2017), and most of the dwellers make living close the slum (Saika & Matsuyuki, 2017). This condition could be observed clearly, where within slums there is a great human potential to support a region's sustainable development policy by involving local communities and civil society (Elrayies, 2016).
The results also show that slum areas are not a concentrated population with low income and/or low education. These results are supported by (Roy et al., 2018) and (Uddin, 2018) where income levels vary among residents in slums. There are land owners and land tenants upon socio-economic system that have been established for a long time in slums, and the arrival of poor individuals who stepped in into this circle has no impact towards overall environment (Duah & Bugri, 2016;Nakamura, 2016). The impression of poverty attached to slums is a result of building density, while slums located in the downtown perform good economic opportunities (Bird et al., 2017).
This study is considered essential to investigate the priority location of slum improvement management, hence it was right on target. Studies based on spatial analysis of physical indicators of building density would complement previous studies conducted based on population censuses, as those studies were able to reduce bias data, where population data were not equivalent to location of residence.

Conclusion
Applying cluster and outlier modeling of building density levels of urban slums in Jakarta based on RT boundary level revealed more specific slum locations and identified factors that influencing differences of building density levels. Slum locations that had direct neighbors to areas designated at Jakarta spatial detail planning as industrial zones and warehousing areas as well as office, trade and service zones, possed a higher level of building density compared to slum locations that had neighbors to areas that were designated as housing zones. This current study showed the level of income and education in each RT area did not significantly providing influence upon slum area. Further studies could be applied by analyzing the level of building density in all areas of Jakarta combined with other variables to obtain more specific slum distribution clusters. This study suggests that DKI Jakarta Provincial Government could determine the more appropriate solution based on the characteristics of each slum area in dealing with slum improvement in Jakarta.

Conflict of Interests
The authors declare that there is no conflict of interest with any financial, personal, or other relationships with other people or organizations related to the material discussed in the article.