Using social media data to predict riding results in the
2019 Canadian federal election
PRESENTER:
Shainen Davidson and Kenton White
Advanced Symbolics Inc.
ABSTRACT:
Using social media data, as well as census, and historical election and
polling data, we trained a model to predict the vote share in each of the
338 ridings in the 2019 Canadian federal election. We will present the
model, and examine how social media data may be used to predict human
behaviour. By the presentation date of this talk the election will have
concluded, so we will also examine the performance of the model: what went
right, and where the model showcased the limits of what we can predict
using these data sources.