My dataframe called ozon with one column O3 is filled with hundreds of series with NaN values. In itself it\'s not a problem, however
ozon
O3
NaN