One of the way by which you can achieve this is by representing categorical variables as binary vectors. Transform categorical variable to a format that works better with classification and regression algorithm. 
Something like this:
| id | 
height | 
weight | 
nationality | 
nationality.us | 
nationality.uk | 
nationality.in | 
nationality.nz | 
nationality.aus | 
| 1 | 
169 | 
53 | 
uk | 
0 | 
1 | 
0 | 
0 | 
0 | 
| 2 | 
145 | 
67 | 
us | 
1 | 
0 | 
0 | 
0 | 
0 | 
| 3 | 
125 | 
45 | 
in | 
0 | 
0 | 
1 | 
0 | 
0 | 
| 4 | 
156 | 
76 | 
aus | 
0 | 
0 | 
0 | 
0 | 
 | 
And then carry on with the usual method.