December 2019 Summaries

The CARTO 5 newsletter has been sharing stories, tips, and tricks on Location Intelligence and Spatial Data Science. The year-end edition highlights the most-clicked articles from the newsletter, including news and new tech, learning resources and scholarship, new data sets, and cool maps. Some of the key points include Google's plans to monetize its Maps app, the release of Facebook's accurate Africa population maps, and the launch of a new satellite data repository called UP42. The newsletter also features showcases of amazing maps and data visualizations, including the Information is Beautiful Awards 2018 winners and Canadian Geographic's best maps of 2018.

Quantifying Park Use in American Cities: Spatial Models and Novel Measures` The traditional methods of evaluating park access using simple spatial models with just distance and area are found to be inaccurate, as they represent potential access rather than actual use. GPS location data now records when smartphones visit parks, allowing for the aggregation of daily park visits to yield usage rates neighborhood by neighborhood. These rates can be compared to the spatial models city by city, revealing that the models perform poorly due to ignoring complex routines of urban life and lacking calibration. The final issue is that access is not simply spatial, even accounting for complexity, and GPS mobility data help make this clear. The analysis reveals that realized access (actual use) is more unequal than potential access, with minority populations having lower levels of park use. GPS data also enable the study of which amenities are actually used within parks, revealing that parking lots are among the most-used features in some parks, and that these spaces can be improved to serve their adopted function.

Transforming spatial data from one scale to another is a challenging task known as a change of support problem (COSP). CARTO's Data Observatory uses quadkeys to create a common grid, which allows for the transformation of data on the finest spatial scale useful for analysis. Statistical methods can be used to enhance the spatial resolution of Census data by selecting a subset of covariates and applying regularization techniques such as LASSO. Transfer learning approaches can also be used to improve model accuracy by adding additional response variables with better predictive skills. The model-based downscaled data can capture spatial clustering and heterogeneity, providing valuable insights for political campaigns and electoral strategy. By using CARTO's statistical downscaling models, users can magnify their analysis and gain a deeper understanding of the relationship between location and business performance.

Tinsa Digital has developed a comprehensive and unbiased dataset of Spanish residential real estate data, which is now available in CARTO's Location Data Stream Catalogue. This dataset provides 100% verified and reliable information on residential properties, including metrics such as average appraised values, rental prices, and mortgage fees. The data is sourced from Tinsa Digital's annual valuations of over 300,000 properties inspected by independent experts. Geospatial analytics platforms have become essential in the real estate sector, allowing for analysis of large portfolios and trends. The inclusion of this dataset in CARTO enables users to perform complex spatial analyses and gain insights into the Spanish real estate market. With this data, professionals can make informed decisions and minimize risk, while also providing a consolidated review of historical indicators.

The coffee industry is one of the most traded agricultural commodities in the world, with 60% of the world's coffee produced by smallholder farmers who face unique challenges in improving production. The true number of smallholder coffee farmers is estimated to be around 12.5 million, with many living in poverty and facing new challenges from a changing climate. Poverty statistics show that nearly half of these farmers live below the international poverty line of $3.20 a day, with the majority concentrated in six East African countries. To address this issue, organizations like Enveritas are using data and geospatial tools to empower smallholder farmers and improve their livelihoods. By recognizing the diversity and magnitude of coffee farmers contributing to our daily cups, we can start to responsibly source coffee and work towards a more sustainable industry.

The text discusses the use of spatial data science techniques, specifically human mobility data and optimization algorithms, to optimize market coverage for a bank expanding into new locations in southern Spain. The goal is to identify the 10 optimal locations that attract the largest number of visitors while minimizing cannibalization between branches. The methodology developed includes data discovery, support selection and cell enrichment with human mobility sociodemographics and commercial data, catchment area calculation, and optimization using an approximation algorithm based on a greedy approach and linear programming. The results show that the algorithm is able to identify 10 locations that cover a large number of potential customers, improving upon a standard greedy algorithm by approximately 11.75%. The methodology can be applied to other sectors facing similar challenges, such as retail, banking, and telco, to make informed decisions about location expansion.