ANALYSIS OF CRIMES IN THE CITY OF NEW YORK

Analysis of Crimes in The City of New York is a Capstone Project developed during Data Science Career Track at Springboard (2018)

It contains 5 files:

1. 'Analysis of Crimes in the City of New York - Capstone Project 1 - Report - Springboard 2018.pdf' - a report describing the project and its outcomes in detail
2.'Analysis of Crimes in The City of New York - Capstone Project 1 - Presentation - Springboard 2018.pdf' - a slide presentation of the report 
3.'Analysis of Crimes in The City of New York  - Capstone Project 1 - Code - Springboard 2018.ipynb' - a code used to develop the outcomes of the report
4 'Analysis of Crimes in The City of New York  - Capstone Project 1 -Demographics Data Wrangling Code - Springboard 2018.ipynb' - a code used to transform demographics data set
5. 'Analysis of Crimes in The City of New York  - Capstone Project 1 - NYPD Data Set Wrangling Code - Springboard 2018.ipynb' - a code used to clean NYPD data

Cleaned Datasets to be run with the code are available at
https://drive.google.com/open?id=1ZIX2W6mE2E5I_EQ_woeBclKOUw_qdyp-


The main sections of the report are:


INTRODUCTION - introduction of the project
DATABASES - description of the databases used during the project
CLIENT - potential client of this project
DATA WRANGLING - wrangling procedures performed on the original dataset
DATA EXPLORATION - exloratory analysis of the datasets
CRIME RATES	- analysis of NYC crime rates
CRIME DENSITY - analysis of crime density 
CRIME RATES AND POPULATION - relationship between crime rates and population
CRIME HOMOGENEITY - comparison of different fractions (area, population, crime rate) among boroughs
CRIME STATUS ANALYSIS - analysis of crime rates of different level
CRIME RATES AND HOUSING MARKET - analysis of relationship between crime rates and housing market in NYC
DATA MODELING (SUPERVISED LEARNING)	- Linear regression model for relationship between severe crime rates and different demographic indicators
DATA MODELING (UNSUPERVISED LEARNING) - study of clustering in the dataset involving: 
	SIMILARITIES BETWEEN BOROUGHS - cosine similarities between boroughs
	PRECINCT SEGMENTATION - methods to investigate clustering among precincts using:
		T-DISTRIBUTED STOCHASTIC NEIGHBOR EMBEDDING	
		PRINCIPAL COMPONENT ANALYSIS	
		CLUSTERING METHODS	
	SIMILARITIES BETWEEN PRECINCTS (TOOL) a tool to find out the top most similar precinct and the most contributing crimes 
ASSUMPTIONS AND LIMITATIONS	- limitations of the analysis
RECOMMENDATIONS AND FUTURE WORK	- recommendation for the next steps
CONCLUSIONS - summary of the analysis