Parking Accessibility Evaluation Tool

Project Info

Team Members

Anita , Daphne , Liqi Wang , des , Swati , Dominic , Beatriz Diaz , Albert

Project Description

Project by GA Data Scientist + Des

CONTEXT: Urban Accessibility can be defined as a measure of effort and resources a person must dedicate to arriving at their destination. The measure is created by calculating the averages of variables such as walking times or other active transport options(bicycle, scooter, skateboard etc.), parking availability, locations and frequency of public transport options and other ride-share opportunities.
An area can be considered as highly accessible when access to essential services meets a designated threshold with consideration to time and distance measures from modes of transportation other than motorised personal vehicles.

Car parking rates are fundamental to address the connectivity of low accessibility zones, as well as mitigate the impact of traffic and pollution in highly connected and accessible areas. Despite this, all across NSW there are large gaps in the recorded data of vehicle parking areas, as well as the prices set and actual usage. Implications of this could be costly development solutions that risk allocating more space than needed in a high accessibility area, or in under providing available parking for locations of low accessibility.

PROBLEM: Despite this, in most cities, there are significant gaps in knowledge around how much car parking spaces are actually used. In NSW, there is currently no reliable way for accessibility to be measured.

SOLUTION: As accessibility increases people tend to have access to more opportunities and connectivity to essential services, transportation services and urban hubs.

Our solution aims to identify low and high accessibility zones that are crucial to empower land use planners with an evidence base system that meets car parking requirements, improves accessibility and mobility, and enable residents to access more opportunities.

What our project has strived to do is analyse the relevant data from ABS and data from the NSW Government to better assess and understand the factors that should be considered when looking at the size of parking spaces.

Sources that we have identified are Current Car parking availability, Population density, Suburb Characteristics such as geography and current road layout, current use of utilities such as public transport and Car Sharing applications, Essential Services locations such as supermarkets and school zones, as well as other modes of active transport such as walking or cycling.

Our dashboards will use all of these factors across each location to help in the identification of High and Low Accessibility areas All across NSW, as well as providing a suggested number of carparks when focused at the suburb level.

All of this data will be utilised and accessible within our created dashboards, which are interactable through the map, making it easy for anyone to locate and understand just how accessibility is as a factor in their suburb.

#nsw #sydney #parking #accessibility #data

Data Story


The following characteristics of a suburb are used to determine the accessibility level:
1. Working age population
2. Number of transport stops (trains, wharves,
bus interchange) in a suburb
3. Population per mode of transport (Suburb level)
4. Number of public parking spaces available in a suburb

We are using a two-step modelling process.

The first step involves applying Machine Learning via BIRCH Clustering, an Unsupervised Learning technique. Through this we are able to create three clusters that will identify accessibility levels for each suburb in our dataset.

Next, the result of the clustering model is used as input to a Logistic Regression Model. The regression model can be used to predict the accessibility level of a suburb that is not in our existing dataset given the suburb’s characteristics.

Looking at the feature importance from the Regression model, we are also able to infer the impact of each suburb characteristics to its accessibility level.

Suburbs that are classified under the low accessibility level cluster exhibit the following characteristics based on the regression model coefficients:
- Likely to have low working age population
- Less likely to have available public parking spaces
- Less likely to have population that use the modes of transport considered in the model
- Most of these suburbs are located outside the urban areas

Suburbs that are classified under the high accessibility level cluster exhibit the following characteristics based on the regression model coefficients:
- More likely to have high working age population
- More likely to have available public parking spaces
- More likely to have population that use the public transport
- Most of these suburbs are located in the major train interchange
- Most of these suburbs are located in the major urban areas

Suburbs classified as medium are those that have characteristics in between the ones mentioned above. These suburbs are also likely to be closer to major urban areas but are likely to have less public parking spaces close to public transport. Hence, people from these suburbs are more likely to use their own vehicles than use public transport. This can be a focus of city planners when aiming to reduce car use by population.

Further improvements are planned to increase the reliability of the model. Some of the issues encountered when implementing the project that need to be addressed are:
Availability of similar data for all suburbs in NSW
We identified that the Public Transport Accessibility Level (PTAL) will be useful data to help improve the model, however, access to this data in a format that can be used is difficult. This will be included in future iteration of the project.
City planning subject matter expertise will help improve the assignment of the clusters and interpretation of drivers that affect classification.

Addition of new features will help us improve the way we determine accessibility scores for the suburbs. The following features will also be considered in the modelling in future iteration:
Car sharing
Essential Services.

The aim of the project is to be able to aid city planners to design better transport systems. Future iterations of the project will include other capabilities such as Scenario-based Planning and Car Park Spaces Demand Forecasting which will leverage the machine learning models.

Evidence of Work



Project Image

Team DataSets

Population by mode of travel to work

Description of Use To understand how the population in the suburb move. This is used in the modelling.

Data Set

Public transport

Description of Use To understand the connectivity of the zone with current transport infrastructure

Data Set

Population Info

Description of Use data used in clustering model

Data Set

Commuter Car Parks

Description of Use To understand the location of parking facilities near to public transport

Data Set

Household Travel Survey

Description of Use To know what methods of transportation are being used

Data Set

Car Share

Description of Use To consider the availability and uptake of this novel transport solutions.

Data Set

Challenge Entries

Getting parking right: matching car parking rates to accessibility

How can we make a difference by making ‘accessibility’ measurable?

Go to Challenge | 4 teams have entered this challenge.