Usage of AI for
analysis of
transit data
By: Team MA02
Dataset Background
- From the many datasets provided in the https://data.gov.au/, we have
considered to pick “Daily Public Transport Passenger Boardings By Passenger
Group”.
- This dataset contains daily Public Transport Patronage in types of passengers
including adults, school students, tertiary, concession, and other in Canberra,
the capital city of Australia. Paper tickets purchased from Ticket Vending
Machines are not included.
Key Insights
● What the dataset is about
● What is the application of AI for this dataset
● How can companies, especially transportation companies, utilise the data
● How the government can plan infrastructure to accommodate future
developments
Demography of Passengers Data
From the large volume of data about the number of daily public transport use by the
population of Canberra, we can obtain:
● Maximum volume of commuters in different age group and category
● Lowest volume of commuters in different age group and category
● Peak days of increasing commuters according to date
● Lowest volume of commuters according to date
● Daily activities affecting the numbers of passenger groups (covid, summer
holidays)
● Yearly Trends
● The distribution of demographic
How do we benefit from the statistics found?
1. Optimizing Service Scheduling and Capacity Planning
2. Targeted Marketing and Discounts
3. Forecasting and Resource Allocation
4. Sustainability and Environmental Impact
5. Policy Development and Public Planning
AI Application
● Reading the data in the most efficient manner
● Can help make predictions by interpolating or extrapolating data
● Health and safety concerns that arise from such traffic of passengers
movement can be tackled
● Advertisement companies can be more cost-effective when developing
marketing campaigns to customers
● Route optimization
● Stimulation and modelling can be used to analyse the impact from road changes
● Security for commuters
Data Abuse Concern
● Data can be stolen and used in illegal activities
● Certain outcomes that were made using the existing data might have different
reasons why they happened, therefore future predictions using new data might
be wrong
● Wrong outcomes because of not filtering the data and/or applying wrong
techniques to analyze it
● Wrong predictions can mislead user to make wrong decisions
● Loss of money due to wrong or imprecise predictions
Data ethics
● Concern: Data can reflect societal biases, leading to discrimination
● Concern: Predictions can be used to change daily routine of the people, which
they may not like
● Solution: Accountability for both AI decisions and human oversight
● Solution: Consider what people think about new implementations
● Solution: Limiting the purpose of data usage (for example, if the data is used to
make larger profit, it’s used only for this purpose)
● Solution: High level of cyber security
Thank
you
