Capstone Project blog

Captsone final REPORT 2020

June 29     Soundqmedia Authored by: PS Ajithkumar

Capstone Project - The Battle of Neighborhoods (Week 2)

 

Report   Table of contents Introduction: Business Problem Data Methodology Analysis Results and Discussion Conclusion

“”

Introduction: Business Problem The aim of this project is to find a safe and secure location for opening of commercial establishments in Vancouver, Canada. Specifically, this report will be targeted to stakeholders interested in opening any business place like Grocery Store in Vancouver City, Canada. The first task would be to choose the safest borough by analysing crime data for opening a grocery store and short listing a neighbourhood, where grocery stores are not amongst the most commom venues, and yet as close to the city as possible. We will make use of our data science tools to analyse data and focus on the safest borough and explore its neighborhoods and the 10 most common venues in each neighborhood so that the best neighborhood where grocery store is not amongst the most common venue can be selected. Data Based on definition of our problem, factors that will influence our decission are: finding the safest borough based on crime statistics finding the most common venues choosing the right neighbourhood within the borough We will be using the geographical coordinates of Vancouver to plot neighbourhoods in a borough that is safe and in the city's vicinity, and finally cluster our neighborhoods and present our findings. Following data sources will be needed to extract/generate the required information: ·         Part 1: Using a real world data set from Kaggle containing the Vancouver Crimes from 2003 to 2019: A dataset consisting of the crime statistics of each Neighbourhoof in Vancouver along with type of crime, recorded year, month and hour. ·         Part 2: Gathering additional information of the list of officially categorized boroughs in Vancouver from Wikipedia.: Borough information will be used to map the existing data where each neighbourhood can be assigned with the right borough. ·         Part 3: Creating a new consolidated dataset of the Neighborhoods, along with their boroughs, crime data and the respective Neighbourhood's co-ordinates.: This data will be fetched using OpenCage Geocoder to find the safest borough and explore the neighbourhood by plotting it on maps using Folium and perform exploratory data analysis. ·         Part 4: Creating a new consolidated dataset of the Neighborhoods, boroughs, and the most common venues and the respective Neighbourhood along with co-ordinates.: This data will be fetched using Four Square API to explore the neighbourhood venues and to apply machine learning algorithm to cluster the neighbourhoods and present the findings by plotting it on maps using Folium. ·           Part 1:  Using a real world data set from  Kaggle containing the Vancouver Crimes from 2003 to 2019 Vancouver Crime Report Properties of the Crime Report TYPE - Crime type YEAR - Recorded year MONTH - Recorded month DAY - Recorded day HOUR - Recorded hour MINUTE - Recorded minute HUNDRED_BLOCK - Recorded block NEIGHBOURHOOD - Recorded neighborhood X - GPS longtitude Y - GPS latitude     Importing all the necessary Libraries   Reading from the Dataset Due to sheer amount of data(~ 600,000 rows), it was not possible to process all of them and instead for this project we will be considering the recent crime report of the 2018.

Changing the name of columns to lowercase

Part 2: Gathering additional information about the Neighborhood from Wikipedia

As part of data set Borough which the neighborhood was part of was not categorized, so we will create a dictionary of Neighbborhood and based on data in the following Wikipedia page.

 

Methodology

Categorized the methodologysection into two parts:

• Exploratory Data Analysis: Visualise the crime repots in different Vancouver boroughs to idenity the safest borough and normalise the neighborhoods of that borough. We will Use the resulting data and find 10 most common venues in beach neighborhood.
•  

• Modelling: To help stakeholders choose the right neighborhood within a borough we will be clusterinvg similar neighborhoods using K - means clustering which is a form of unsupervised machine learning algorithm that clusters data based on predefined cluster size. We will use K-Means clustering to address this problem so as to group data based on existing venues which will help in the decision making process.

 

Results and Discussion

CodeText

---

The objective of the business problem was to help stakeholders identify one of the safest borough in Vancouver, and an appropriate neighborhood within the borough to set up a commercial establishment especially a Grocery store. This has been achieved by first making use of Vancouver crime data to identify a safe borugh with considerable number of neighborhood for any business to be viable. After selecting the borough it was imperative to choose the right neighborhosod where grocery shops were not among venues in a close proximity to each other. We achieved this by grouping the neighborhoods into clusters to assist the stakeholders by providing them with relavent data about venues and safety of a given neighborhood.

 

 

Conclusion

CodeText

---

We have explored the crime data to understand different types of crimes in all neighborhoods of Vancouver and later categorized them into different boroughs, this helped us group the neighborhoods into boroughs and choose the safest borough first. Once we confirmed the borough the number of neighborhoods for consideration also comes down, we further shortlist the neighborhoods based on the common venues, to choose a neighborhood which best suits the business problem.