GIS

1
GIS Retail Site Selection
Paul Amos The Wharton School of Business Wharton
GIS Lab
2
What is Geographic Information System (GIS)?

• Desktop GIS goes beyond desktop mapping systems
  by providing analysis and data manipulation
  tools.
• GIS provides Query and Analysis tools to
  investigate geographic relationships in your data
  and define relationships among multiple sets of
  data.
• Allows Licenses to edit both geographic and
  tabular data. Changes in your data are
  automatically reflected in your maps.

3
(No Transcript)
4
What can you do with GIS?

• Locate your customers and competitors on a map.
• Decide where to set up a store, business, or
  shopping mall.
• Analyze demographics and create store profiles.
• Define trade areas and target customers.
• Target specific areas for Direct Mail campaigns
• Customize the user interface to automate
  repetitive tasks.
• Display Aerial Photographs

5
Locating Customers or Competitors
6
Network Analysis
7
Trade Area Analysis

• Simple Ring Analysis
• Data Driven Analysis
• Equal Competition Analysis
• Drive Time/Distance Analysis
• Customer Analysis

8
Simple Ring Analysis
9
Data Driven Analysis
10
Equal Competition Analysis
11
Drive Time/Distance Analysis
12
Customer Analysis
13
Aerial Photographs
14
Project Objectives

• Develop Econometric Sales Forecast
• Model
• Create a Site Selection Decision Support
  
• Tool
• - Accessible to a Centralized Data
  Warehouse
• Customized GIS Front End

15
Independent Variables
Demographic - Population - Income
- Housing - Growth Rates

• Site Location Data
• - Store Size
• - Traffic Count
• - Proximity to other units
• - Real Estate Costs

Business Climate - Competition -
Businesses by Type - Retail Concentration
- Daytime Population

• HR and Operations
• - Staffing and Training
• - Marketing and Advertising

16
Model Development

• We created interaction, transformed and
  dummy variables
• Assigned weighting schemes to trade area
  variables
• Regressed the independent variables against
  the dependent variable, Sales
• Used SAS to perform the regression analysis

17
Model Output
Sales Square footage 1,000 HHs 2,500
Number of Competitors -250 ...
18
First Year Model Results

• 26 Variables were found to be
  Statistically Significant
• R2 0.74
• Average Predicted vs. Observed Sales
  Error 10
• 80 of all Predictions are within 16 of
  Actual Store Sales

19
Model Enhancements

• Trade Area Computation - Centroid
  Approach vs. Proportional Area
• Traffic Counts
• Store Information

20
Model Results
21
Create Data Warehouse

• Database Management System - Oracle
• Demographic Data at Block Group Level
• Competition by SIC Code
• Crime Information at Census Tract Level
• Store Spatial Data - Maps

22
GIS Software Application

• Database Security Login
• Locate Prospective Site
• Create Trade Areas for the Site
• Analyze Demographics, Competition, Crime
  Information
• Generate a Sales Forecast
• Output Reports

23
Database Security
24
Locate a Prospective SiteAutomatically
25
Automatic Site Search
26
State Map
27
Locate a Prospective SiteManually
28
Manual Site Search
29
Create Trade Areas
30
Create Trade Areas
31
Demographic Analysis
32
Competition Analysis
33
Crime Analysis
34
Generate Sales Forecast
35
Select Reports
36
Sample Report
37
Online Reporting Subscriptions

• Annual Subscriptions - 995 per user
• Anysite
• Datametrix
• DemographicsNow.com
• Mapscape.com
• Sites USA
• Additional Subscriptions
• Crime Risk
• Traffic Counts
• Lifestyle Clusters
• Weather
• Earthquake Information

38
Data Costs

• Demographic Data
• Regional Coverage - 12,000 (Unlimited Licenses)
• Crime Data - 25,000 (Unlimited Licenses)
• Map Data
• Regional Coverage - 16,000 (Unlimited Licenses)
• Software
• Database - 25,000 (8 Licenses)
• GIS Software - 10,000 (10 Licenses)
• Address Matching Software - 7,000 (10 Licenses)
• Total Startup Cost - 95,000

39
Benefits

• The Product Pays for Itself
• Takes a Scientific Approach to Real Estate
• Improves Decisionmaking
• The Product is Scalable
• Data Sales - Custom Databases
• Sales Forecasting Methodology
• Can be developed in any Object Oriented Language
• CD, PC, Client-Server or the Internet

40
Benefits

• Sales Forecast Model created from an
  independent, unbiased Source
• Time - Sites can be processed faster
• Money - Data costs decrease per site
• Proactively vs. Reactive searching for new
  Sites
• The Application can be accessed Remotely with a
  Laptop and Modem

41
Contact Information

• Paul Amos
• Executive Director
• Wharton GIS Lab
• 3401 Market Street
• Suite 103
• [phone]
• amosp_at_wharton.upenn.edu