TaxLess: Optimising your tax return

Team Name: 
TeamTax

TaxLess is a simple to use app that will help you with your next tax return. TaxLess optimises your next tax return by comparing your deductions with people like you.

We took open ATO dataset, applied some predictive analytics to understand which socio-economic catergory you fall under and created web tool that discovers where your deductions are falling short and what can you do about them.

How we do it

Initially, TaxLess asks you to complete a short form about yourself and your income. Alternitavely, you can upload your tax return from the previous year and we will automatically fill this form for your convenience.  By using Machine Learning algorithms, we group you with similar taxpayers, and find out how you rank in the various deduction categories among them.

A significant difference likely indicates that you are either under claiming or under investing. For example, you may be self-contributing nothing to supperannuation when others are doing 10% of their income, or you simply forgot to report expenses. 

Going beyond your tax return

Our aim is to empower users with no statistical or finance background to understand how tax return works and ultimately to help them to claim optimal amount of money. 

Used Datasets: 
Dataset Name: 
Taxation statistics - Individual tax return sample files
Publishing Organisation/Agency: 
ATO
Jurisdiction of Data: 
Australian Government
How did you use this data in your entry?: 
Firstly, we use K-means clustering (by Occupation code, salary and wages amount, region, age, marital status and gender) to group users into clusters. Secondly, within the clusters we use ridge regression (aided with dummy variables for the discrete variables) to understand, in a fine grained way, how tax effective particular users are and where they get deductions. Finally, we profile the clustering by showing the distribution of variables in the different groups to study their differences among the clusters.
Event Location: 
Sydney Official