07-conclusion.Rmd

# Conclusions and Future Directions

This work balances quantitative and qualitative views on our urban world at the site scale, where incremental urban change occurs. We believe its cross-disciplinary nature, along with focus on a values-driven application for quantitative analysis, holds great potential for future practice in the development of our built world, which must balance disciplinary intuition and values with computational capacity and increasingly large data sets through carefully designed processes for human-machine collaboration.

The method of segregating quantitative and qualitative analysis was highly productive. The factor-based analysis yielded intriguing results that require further consideration and mediation through the inclusion of expert values-based input. While some variables tested held no relevance at the site level in relationship to factor analysis, we believe there are other variables that should be included in the future that would optimize analysis in relationship to on-the-ground experience of urban quality and sustainability. Overall, we find that site-level analysis is complimentary to city- and neighborhood-scale analysis in that it provides a clear picture of the “natural” boundaries and fault lines of a place rather than those drawn by governmental authority. This scale also allows for an understanding of edge conditions and block-scale nuance, which strongly shape the experience of a place.

In more detail, we find that a combination of buffering and discreet site-level data is a promising pathway toward an understanding of the qualities of an area at the site level, and that such methods should pursued in the future with different data types. For example, there is an unexplored opportunity to integrate climate-related factors in a more robust way. Right now, the measure of environmental sustainability is contingent upon carbon production in relationship to transportation. In the future, parcel level metrics including urban heat island effect (via Landsat data) and building type and age (via parcel data) may hold potential for exploration. In alignment with the ambition pursuing parcel-level analysis that closer approximates human perception of a place, we have also begun to explore the possibility of systematically engaging street level views of each parcel utilizing Google Street Views and computer vision. The addition of such data, which is a close approximation of human experience on the ground, may enable this research to better bridge the intents of planning and design: to better locate or prioritize the location of future housing within an urban area, and to recommend relevant unit and building types for a given site and its local context.

While initial ambitions for this parcel-level analysis and visualization relate specifically to small-scale (2-4 unit) infill housing, we are aware that there are other relevant applications and stakeholders. Future work may focus on transportation planning or use-related zoning optimization. Applications that focus more on recommendations for unit or housing type would also be useful.
For our research team, future work should prioritize 1) refinement of the data sets and quantitative analysis based upon the inclusion of an approximation of human perspective at the site level, 2) deeper exploration of variables related to building-level sustainability, 3) collection of additional workshop related data from experts in real estate development and community advocacy, and 4) pursuit of a web application for data visualization that utilizes the values-driven perspectives of disciplinary experts to drive the design of an interface layer between human stakeholders and quantitative analysis at the site level.

```{r fig7a, echo = FALSE, out.width='100%', fig.cap='Screen captures of the web application demonstrate early tests that translate quantitative analysis into an interactive application to visualize urban quality at the site level. The images show how the indices can be weighted by the user to reflect user values and needs.'}
here("04_figures",
     "fig_7a.jpg") %>%
knitr::include_graphics()
```

```{r fig7b, echo = FALSE, out.width='100%', fig.cap='Early interaction design tests with a two different versions of user controls that weight the indices to enable the user to personalize their visualization of the research results and their understanding of the site-level qualities of the study area. This function allows users to self-weight the data according to their individual needs. Right now we have directly translated the results of quantitative analysis into the application. In the future, the label of the indices will change for clarity and to align with disciplinary values.'}
here("04_figures",
     "fig_7b.jpg") %>%
knitr::include_graphics()
```

Such a tool would enable stakeholders to access and calibrate indices based upon essential expert values alongside the individual stakeholder value systems and information needs. We have already begun this work and have created a web application that enables users to calibrate the visualization of the data presented above through the lens of the quantitative indices. Progress on this aspect of the work can be explored at this link: https://leahwelch.github.io/projects/pittsburgh-map-sm/. The optimization of this substantial data set for web visualization is a work in progress and we are currently working to employ a better strategy for both virtual machine use and legible interactivity. Additional consideration around the meaning of the factors vis a vis disciplinary values (the creation of a just city, for instance) is needed in order for such a tool to optimize relevance in practice with the aim of better rather than (simply) more development.