Identified slow parts of the code

[abelsiqueira]

Some identified slow parts of the code:

• radar_dataset.py:71 - Super constructor. I guess there is nothing we can do? Maybe it is just an outlier of the file that I am testing.
• radar_dataset.py:108 - Comparison of a sparse matrix with a number: distance_matrix > 0. This distance_matrix has dimensions (number of rows in all data)x(number of not NA rows), so it has to be sparse (any distance over max_distance is 0).
• radar_dataset.py:115 - indexing data.loc[indexes, xyz] in the creation of local_tree.
• radar_dataset.py:122 - indexing in data.loc[indexes, self.features].fillna(0).
• radar_dataset.py:123 - any on assertion - which shouldn't be there any more

Other slow operations are import pandas as np, read_csv.
Not so slow operations: "convert (rows,cols) to create graph", save_graphs,