Exploratory Spatial Analysis
We used the PySAL package to quantify the geographic information, getting the adjacent weights and lag spatial weights of features from the GeoDataframe. We map the established size to five classes and compare the difference between weighted establishment size and the original one.
User Reviews Spatial Analysis
We used user review datasets to cluster and aggregate the distribution of different types of small businesses, average rating, high rating (score >=4.5) distribution, review counts, sentiment scores distribution (fig2) within each geo id in Portland.
23:'Construction'
42:'Wholesale Trade'
54:'Professional, Scientific, and Technical Services'
62:'Health Care and Social Assistance'
72:'Accommodation and Food Services'
81:'Other Services'
Example of sentiment socres in different GeoID
We also calculate sentiment scores among different types in different GeoID(fig3). From these analyses, we derive an understanding of current small business conditions and feedback. We have also used a version of these maps on the final product to support our recommendations.
Conclusion
Decision Support Systems for locations is a helpful tool in the hands of small-business owners since it gives an opportunity to access information on the level of multinational companies. Organizations have long resorted to scoring mechanisms to quantitatively compare against locations in viable markets. In this project, we have aggregated data from multiple sources to come up with a score that ranks Geo-IDs based on said score, similar to that used by large organizations. Although we have tried to emulate the process done by larger organizations, we have also tried to tailor the approach to suit the needs of small businesses. It has to be pointed out that these results are not final and error-free, with further research and effort this product can become a reliable go-to option.
​