More New Yorkers are living on the streets and homeless shelters than ever before. There are thousands of homeless people sleeping in the shelters and on the streets. We are trying to figure out solutions that can help Department of Homeless Services (DHS) tackle this homelessness problem. Our research consists of three parts. We used data from 311 datasets, datasets provided by the DHS, NYPD, and the demographics dataset to create a supervised learning model which would predict the hot and cold spots in NYC to find homeless people. We also evaluated the effectiveness of the homeless drop-in centers using hypothesis testing and found that the drop-in centers may not be that effective. We analyzed the homeless data and discovered evidence on how policies affect the homeless population.
Vishwajeet Shelar (V.S) drafted the major part of the report. Chuan-Heng(Henry) Lin (C.H.L), Chunqing Xu (C.X) and Dongjie Fan (D.F) helped in drafting the methodology.
V.S, D.F, C.H.L and C.X collected datasets and cleaned the data individually. C.X cleaned the homeless shelter data and plotted the homeless shelter population time-series. V.S plotted and analysed the time-series (decomposition) of the 311 complaints. C.X and V.S worked together in analyzing the patterns in the time-series.
V.S and D.F worked on the hypothesis testing, D.F formulated the hypothesis test and implemented it. V.S contributed to the interpretation and conclusion.
D.F, C.X completed the spatial analysis of the 311 complaints data. C.H.L, V.S and C.X worked together to create feature space for the model. C.H.L implemented the models. D.F, C.H.L, V.S and C.X worked together to interpret the results of the models