|
1
|
|
|
from . import df_operations as dfop |
|
2
|
|
|
from green_magic.data.base_handling import BaseDataHandler |
|
3
|
|
|
|
|
4
|
|
|
|
|
5
|
|
|
@BaseDataHandler.register_as_subclass('df-handler') |
|
6
|
|
|
class DataFHandler(BaseDataHandler): |
|
7
|
|
|
|
|
8
|
|
|
def get_all_variables(self, *args, **kwargs): |
|
9
|
|
|
return args[0].columns |
|
10
|
|
|
|
|
11
|
|
|
def get_categorical_variables(self, *args, **kwargs): |
|
12
|
|
|
return dfop.categorical_feats(args[0]) |
|
13
|
|
|
|
|
14
|
|
|
def get_numerical_variables(self, *args, **kwargs): |
|
15
|
|
|
return dfop.numerical_feats(args[0]) |
|
16
|
|
|
|
|
17
|
|
|
def missing(self, df): |
|
18
|
|
|
return dfop.missing_values(df) |
|
19
|
|
|
|
|
20
|
|
|
def add_state(self, dataset, feature, computer, state, cache_prev=True, **kwargs): |
|
21
|
|
|
values = computer(dataset, feature) |
|
22
|
|
|
prev_state = feature.state |
|
23
|
|
|
feature.update(state.key, state.reporter) |
|
24
|
|
|
dataset.datapoints.observations[state.index] = values |
|
25
|
|
|
if not cache_prev: |
|
26
|
|
|
self.del_state(dataset, feature, prev_state) |
|
27
|
|
|
|
|
28
|
|
|
def del_state(self, dataset, feature, state, **kwargs): |
|
29
|
|
|
# DEBUG CODE |
|
30
|
|
|
if state.key == feature.current: |
|
31
|
|
|
raise RuntimeError(f"Requested to delete attribute/column '{state.key}', but it is the current state of feature {str(feature)}") |
|
32
|
|
|
del dataset.datapoints.observations.df[f'{feature.id}-{state.key}'] |
|
33
|
|
|
del feature.states[state.key] |
|
34
|
|
|
|
|
35
|
|
|
|
|
36
|
|
|
data_handler = DataFHandler() |
|
37
|
|
|
|
|
38
|
|
|
|
boromir674 /
so-magic
