DEV Community

Discussion on: Neural Networks are Overrated

Collapse
andimarafioti profile image
Andres Marafioti

Hey, nice article. I was wondering, how did you calculate the baseline?

Collapse
mobbsdev profile image
Ian Mobbs Author

Hi Andres,

The baseline is the accuracy if you just guessed the most common result for each classification. So in this example, I believe 'White' was the most common race in the dataset at around 49% of the samples. So if you just classified every race as 'White', you'd have a 49% accuracy - hence, it's the baseline!