Multivariate Predictive Framework for ATM Placement Using Census, Social Media, and Open Data Sources
Publication Date: 2016-Nov-15
The IP.com Prior Art Database
Disclosed are a method and model to enable financial organizations (i.e., banks) to measure expected Automated Teller Machine (ATM) usage in certain areas before placing ATMs, which allows banks to identify the area(s) most beneficial to the ATM strategy. The quantitative framework gathers data from open data sources, applies four regression modeling techniques, and utilizes transaction data from a bank's ATM assets to perform an analysis and identify areas for successful ATM placement.
Page 01 of 5
Multivariate Predictive Framework for ATM Placement Using Census , Social Media, and Open Data Sources
As Automated Teller Machine (ATM) revenue is a billion-dollar business, optimizing ATM placement is of high interest to banks. Identifying the optimal location to place an ATM is a problem yet unsolved for banks worldwide.
Understanding effective placement of an ATM requires some measurement of the correlation between ATM placement in certain areas and the ATM's usage; therefore, a quantitative vetting process is needed to identify effective ATM placement. The goal is to identify areas where ATMs have a high likelihood of repeated and consistent use. A bank can then place ATMs in areas with high human foot traffic to capture the greatest number of potential customers, but away from areas already saturated with competitors' ATMs. Once a few such areas have been identified, use a method of data collection, transformation, and statistical regression to predict the most valuable areas.
The novel method and model enable banks to measure expected ATM usage in certain areas before placing ATMs, which allows banks to identify the area(s) most beneficial to the ATM strategy. Current methods rely on trial and error, which can waste operational dollars.
The novel solution offers a quantitative framework for vetting ATM placement. The core idea is that local ATM cash flow is directly dependent upon the local demographic characteristics, socio-economic factors, and business density immediately surrounding ATMs. To help banks identify favorable areas of a city in which to place ATMs, the novel model uses all of the following:
Several open data sources (e.g., census, social media, geo-specific data via Google* Application Programming Interfaces (APIs))
Four regression modeling techniques
Transaction data from a bank's ATM assets
Thus, using publicly available data, the model and method create a data set that can describe a broad range of local neighborhood characteristics and associate these attributes with ATM usage on an extremely local level. It measures all characteristics of a micro geographic sub-region (e.g., Census Tract) and augments this data set with ATM usage at the Tract level.
Figure 1: Spotting areas suitable for new ATM placement
Page 02 of 5
The novel model predicts ATM usage given characteristics of the ATM's immediate geography (Figure 1) that are gathered from open data sources (e.g., census, social media, Google APIs, etc.). This model is best used within a framework for ATM placement that us...