Pandas usa mapa lambda a fillna python
import numpy as np
df1['user_score'] = np.where(df1['user_score'].isna(), df2['user_score'], df1['user_score'])
Joyous Jaguar
import numpy as np
df1['user_score'] = np.where(df1['user_score'].isna(), df2['user_score'], df1['user_score'])