We have developed a web-based, self-improving and overfitting-resistant automated machine learning tool tailored specifically for liquid biopsy data, where machine learning models can be built without the user's input.

We examine the effect of the Philadelphia LandCare vacant lot greening initiative on crime and the extent to which surrounding land uses and business types moderate this intervention. We rely on a propensity score matching analysis to account for substantial differences in demographic, economic, land use, and business characteristics between greened and ungreened vacant lots. We estimate larger and more significant crime reductions around vacant lots that are greened in our matched pairs analysis compared to unmatched analyses. The effects of vacant lot greening on crime are larger in areas with high residential, high civic, and low transportation land use and are moderated by the presence of different types of nearby businesses.