“The method (statistical vs. ml) should be determined based on the data size and feature distribution, using an ensemble of both is also possible.” @aydinhan@VNGRSAI https://t.co/H8qx89s2ql
Excellent empirical paper co-authored by forecasting luminary Makridakis (@spyrosmakrid) on Machine Learning prediction - results suggest #ml is still in its infancy with lower relative performance to statistical & naive methods but with great potential
https://t.co/G6TfcceM9i