Mankind Vs Machine Sydney Prediction Challenge
Introduction
The
Mankind vs Machine
competition requires data for either bespoke or automated predictions to be compared.
I decided it would be great if this data and competition could tie into a broader social purpose, and introduce everybody to an awesome open dataset and a compelling problem.
THE OPEN DATASET:
The Australian Charities and Not for Profit Commission (ACNC) Annual Information Statement, containing activity, beneficiary, and financial information for all registered charities in Australia. Data has been combined over 2014, 2015, and 2016.
THE PROBLEM:
Optimisation of government grants to the non profit sector (which types of services need it most)?
A POSSIBLE SOLUTION:
Creating an estimation of a given charity's income from donations/bequests/etc.
THE TARGET FEATURE:
[donations_and_bequests]
, which is the charity's reported income from donations or bequests for a given fiscal year.
THE EVALUATION METRIC:
Root Mean Square Error
on the log(1+x) transformed prediction, where x is the predicted target feature
[donations_and_bequests]
THE PRIZE:
GLORY
and Swag from GA
More about this competition, and an intro to the AutoML service TPOT, can be found using these links. Over time, more and more information will be available in the competition Kernals.
Intro Talk:
http://bit.ly/2SVkFpw
Github repo with TPOT example:
http://bit.ly/2qAwaWz
Acknowledgements
We thank the ACNC for making this dataset available!