Skip to content

Commit

Permalink
#1031
Browse files Browse the repository at this point in the history
  • Loading branch information
@steveyuwono
steveyuwono committed Jul 17, 2024
1 parent e1bd4fb commit 12861b9
Showing 1 changed file with 58 additions and 34 deletions.
92 changes: 58 additions & 34 deletions src/mlpro/bf/streams/tasks/deriver.py
View file
Original file line number Diff line number Diff line change
Expand Up @@ -9,10 +9,12 @@
## -- 2023-02-05 1.0.0 SY First version release
## -- 2024-05-10 1.0.1 DA/SY Bugfix in Deriver.__init__()
## -- 2024-05-22 1.1.0 DA Refactoring
## -- 2024-07-17 1.1.1 SY Method Deriver._prepare_derivation(): takeover of feature
## -- and label space from first instance
## -------------------------------------------------------------------------------------------------

"""
Ver. 1.1.0 (2024-05-22)
Ver. 1.1.1 (2024-07-17)
This module provides a stream task class Deriver to derive the data of instances.
"""
Expand All @@ -21,7 +23,7 @@
from mlpro.bf.exceptions import *
from mlpro.bf.various import Log
from mlpro.bf.mt import Task
from mlpro.bf.math import Set, Element
from mlpro.bf.math import Element
from mlpro.bf.streams import Instance, InstDict, StreamTask, Feature, Label
from mlpro.bf.physics import TransferFunction
import numpy as np
Expand All @@ -47,11 +49,7 @@ class Deriver(StreamTask):
p_visualize : bool
Boolean switch for visualisation. Default = False.
p_logging
Log level (see constants of class Log). Default: Log.C_LOG_ALL
p_features : list
The list of current features in the stream. Default = None.
p_labels : list
The list of current labels in the stream. Default = None.
Log level (see constants of class Log). Default: Log.C_LOG_ALL.
p_derived_feature : Feature
A pre-selected feature that would like to be derived. Default = None.
p_derived_label : Feature
Expand All @@ -71,9 +69,7 @@ def __init__( self,
p_range_max = Task.C_RANGE_THREAD,
p_duplicate_data:bool = False,
p_visualize:bool = False,
p_logging = Log.C_LOG_ALL,
p_features:list = None,
p_labels:list = None,
p_logging = Log.C_LOG_ALL,
p_derived_feature:Feature = None,
p_derived_label:Label = None,
p_order_derivative:int = 1,
Expand All @@ -92,29 +88,16 @@ def __init__( self,
if not isinstance(p_derived_feature, Feature):
raise ParamError('Please provide a feature to be derived')

self._idx_feature = p_features.index(p_derived_feature)
self._feature_space = Set()
self._order_derivative = p_order_derivative
self._tic = 0
self._mem_values = []
self._mem_time_stamp = []

for feature in p_features:
self._feature_space.add_dim(p_dim=feature)
feature = p_derived_feature.copy()
feature._name_short = feature._name_short + ' OD-' + str(self._order_derivative)
feature._name_long = feature._name_long + ' OD-' + str(self._order_derivative)
self._feature_space.add_dim(p_dim=feature)

if (p_labels is not None) and (not isinstance(p_derived_label, Label)):
self._idx_label = p_labels.index(p_derived_label)
self._label_space = Set()
for label in p_labels:
self._label_space.add_dim(p_dim=label)
label = p_derived_label.copy()
label._name_short = label._name_short + ' OD-' + str(self._order_derivative)
label._name_long = label._name_long + ' OD-' + str(self._order_derivative)
self._label_space.add_dim(p_dim=label)
self._order_derivative = p_order_derivative
self._tic = 0
self._mem_values = []
self._mem_time_stamp = []

self._derived_feature = p_derived_feature
self._derived_label = p_derived_label
self._feature_space = None
self._label_space = None
self._prepared = False

self._derivative_func = DerivativeFunction(p_name='derivative_func',
p_type=TransferFunction.C_TRF_FUNC_CUSTOM,
Expand All @@ -123,6 +106,39 @@ def __init__( self,
order=self._order_derivative)


## -------------------------------------------------------------------------------------------------
def _prepare_derivation(self, p_inst:Instance):

# 1 Feature space
features = p_inst.get_feature_data().get_dim_ids()
self._feature_space = type(p_inst.get_feature_data().get_related_set())()
self._idx_feature = features.index(self._derived_feature.get_id())

for feature in p_inst.get_feature_data().get_related_set().get_dims():
self._feature_space.add_dim(p_dim=feature)

feature = self._derived_feature.copy()
feature._name_short = feature._name_short + ' OD-' + str(self._order_derivative)
feature._name_long = feature._name_long + ' OD-' + str(self._order_derivative)
self._feature_space.add_dim(p_dim=feature)

# 2 Label space
try:
labels = p_inst.get_label_data().get_dim_ids()
self._label_space = type(p_inst.get_label_data().get_related_set())()
self._idx_label = labels.index(self._derived_label.get_id())

for label in p_inst.get_label_data().get_related_set().get_dims():
self._label_space.add_dim(p_dim=label)

label = self._derived_label.copy()
label._name_short = label._name_short + ' OD-' + str(self._order_derivative)
label._name_long = label._name_long + ' OD-' + str(self._order_derivative)
self._label_space.add_dim(p_dim=label)
except:
labels = []


## -------------------------------------------------------------------------------------------------
def _derive_data(self, p_inst:Instance):

Expand Down Expand Up @@ -175,7 +191,15 @@ def _derive_data(self, p_inst:Instance):


## -------------------------------------------------------------------------------------------------
def _run(self, p_inst : InstDict ):
def _run(self, p_inst:InstDict):

if not self._prepared:
try:
(inst_type, inst) = next(iter(p_inst.values()))
self._prepare_derivation(p_inst=inst)
self._prepared = True
except:
return

for (inst_type,inst) in sorted(p_inst.values()):
self._derive_data(p_inst=inst)
Expand Down

0 comments on commit 12861b9

Please sign in to comment.