Machine Learning with Swift

One-hot encoding

Most of the machine learning algorithms can't work with the categorical variables, so usually we want to convert them to the one-hot vectors (statisticians prefer to call them dummy variables). Let's convert first, and then I will explain what this is:

In []: 
features = pd.get_dummies(features, columns = ['color']) 
features.head() 
Out[]: 

length	fluffy	color_light black	color_pink gold	color_purple polka-dot	color_space gray
0	27.545139	True	0	1	0	0
1	12.147357	False	0	1	0	0
2	23.454173	True	1	0	0	0
3	29.956698	True	0	1	0	0
4	34.884065	True	1	0	0	0

 

So now, instead of one column, color, we have four columns: color_light black ...