You can not select more than 25 topics
Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
148 lines
6.5 KiB
148 lines
6.5 KiB
4 years ago
|
from models import Model, ModelState, AnalyticSegment, ModelType
|
||
|
|
||
|
from analytic_types import TimeSeries
|
||
|
from analytic_types.learning_info import LearningInfo
|
||
|
|
||
|
from scipy.fftpack import fft
|
||
|
from typing import Optional, List
|
||
|
from enum import Enum
|
||
|
import scipy.signal
|
||
|
import utils
|
||
|
import utils.meta
|
||
|
import pandas as pd
|
||
|
import numpy as np
|
||
|
import operator
|
||
|
|
||
|
POSITIVE_SEGMENT_MEASUREMENT_ERROR = 0.2
|
||
|
NEGATIVE_SEGMENT_MEASUREMENT_ERROR = 0.02
|
||
|
|
||
|
@utils.meta.JSONClass
|
||
|
class StairModelState(ModelState):
|
||
|
|
||
|
def __init__(
|
||
|
self,
|
||
|
confidence: float = 0,
|
||
|
stair_height: float = 0,
|
||
|
stair_length: float = 0,
|
||
|
**kwargs
|
||
|
):
|
||
|
super().__init__(**kwargs)
|
||
|
self.confidence = confidence
|
||
|
self.stair_height = stair_height
|
||
|
self.stair_length = stair_length
|
||
|
|
||
|
|
||
|
class StairModel(Model):
|
||
|
|
||
|
def get_state(self, cache: Optional[dict] = None) -> StairModelState:
|
||
|
return StairModelState.from_json(cache)
|
||
|
|
||
|
def get_stair_indexes(self, data: pd.Series, height: float, length: int) -> List[int]:
|
||
|
"""Get list of start stair segment indexes.
|
||
|
|
||
|
Keyword arguments:
|
||
|
data -- data, that contains stair (jump or drop) segments
|
||
|
length -- maximum count of values in the stair
|
||
|
height -- the difference between stair max_line and min_line(see utils.find_parameters)
|
||
|
"""
|
||
|
indexes = []
|
||
|
for i in range(len(data) - length - 1):
|
||
|
is_stair = self.is_stair_in_segment(data.values[i:i + length + 1], height)
|
||
|
if is_stair == True:
|
||
|
indexes.append(i)
|
||
|
return indexes
|
||
|
|
||
|
def is_stair_in_segment(self, segment: np.ndarray, height: float) -> bool:
|
||
|
if len(segment) < 2:
|
||
|
return False
|
||
|
comparison_operator = operator.ge
|
||
|
if self.get_model_type() == ModelType.DROP:
|
||
|
comparison_operator = operator.le
|
||
|
height = -height
|
||
|
return comparison_operator(max(segment[1:]), segment[0] + height)
|
||
|
|
||
|
def find_segment_center(self, dataframe: pd.DataFrame, start: int, end: int) -> int:
|
||
|
data = dataframe['value']
|
||
|
segment = data[start: end]
|
||
|
segment_center_index = utils.find_pattern_center(segment, start, self.get_model_type().value)
|
||
|
return segment_center_index
|
||
|
|
||
|
def do_fit(
|
||
|
self,
|
||
|
dataframe: pd.DataFrame,
|
||
|
labeled_segments: List[AnalyticSegment],
|
||
|
deleted_segments: List[AnalyticSegment],
|
||
|
learning_info: LearningInfo
|
||
|
) -> None:
|
||
|
data = utils.cut_dataframe(dataframe)
|
||
|
data = data['value']
|
||
|
window_size = self.state.window_size
|
||
|
last_pattern_center = self.state.pattern_center
|
||
|
self.state.pattern_center = utils.remove_duplicates_and_sort(last_pattern_center + learning_info.segment_center_list)
|
||
|
self.state.pattern_model = utils.get_av_model(learning_info.patterns_list)
|
||
|
convolve_list = utils.get_convolve(self.state.pattern_center, self.state.pattern_model, data, window_size)
|
||
|
correlation_list = utils.get_correlation(self.state.pattern_center, self.state.pattern_model, data, window_size)
|
||
|
height_list = learning_info.patterns_value
|
||
|
|
||
|
del_conv_list = []
|
||
|
delete_pattern_timestamp = []
|
||
|
for segment in deleted_segments:
|
||
|
segment_cent_index = segment.center_index
|
||
|
delete_pattern_timestamp.append(segment.pattern_timestamp)
|
||
|
deleted_stair = utils.get_interval(data, segment_cent_index, window_size)
|
||
|
deleted_stair = utils.subtract_min_without_nan(deleted_stair)
|
||
|
del_conv_stair = scipy.signal.fftconvolve(deleted_stair, self.state.pattern_model)
|
||
|
if len(del_conv_stair) > 0:
|
||
|
del_conv_list.append(max(del_conv_stair))
|
||
|
|
||
|
self._update_fitting_result(self.state, learning_info.confidence, convolve_list, del_conv_list)
|
||
|
self.state.stair_height = int(min(learning_info.pattern_height, default = 1))
|
||
|
self.state.stair_length = int(max(learning_info.pattern_width, default = 1))
|
||
|
|
||
|
def do_detect(self, dataframe: pd.DataFrame) -> TimeSeries:
|
||
|
data = utils.cut_dataframe(dataframe)
|
||
|
data = data['value']
|
||
|
possible_stairs = self.get_stair_indexes(data, self.state.stair_height, self.state.stair_length + 1)
|
||
|
result = self.__filter_detection(possible_stairs, data)
|
||
|
return [(val - 1, val + 1) for val in result]
|
||
|
|
||
|
def __filter_detection(self, segments_indexes: List[int], data: list):
|
||
|
delete_list = []
|
||
|
variance_error = self.state.window_size
|
||
|
close_segments = utils.close_filtering(segments_indexes, variance_error)
|
||
|
segments_indexes = utils.best_pattern(close_segments, data, self.get_extremum_type().value)
|
||
|
if len(segments_indexes) == 0 or len(self.state.pattern_center) == 0:
|
||
|
return []
|
||
|
pattern_data = self.state.pattern_model
|
||
|
for segment_index in segments_indexes:
|
||
|
if segment_index <= self.state.window_size or segment_index >= (len(data) - self.state.window_size):
|
||
|
delete_list.append(segment_index)
|
||
|
continue
|
||
|
convol_data = utils.get_interval(data, segment_index, self.state.window_size)
|
||
|
percent_of_nans = convol_data.isnull().sum() / len(convol_data)
|
||
|
if len(convol_data) == 0 or percent_of_nans > 0.5:
|
||
|
delete_list.append(segment_index)
|
||
|
continue
|
||
|
elif 0 < percent_of_nans <= 0.5:
|
||
|
nan_list = utils.find_nan_indexes(convol_data)
|
||
|
convol_data = utils.nan_to_zero(convol_data, nan_list)
|
||
|
pattern_data = utils.nan_to_zero(pattern_data, nan_list)
|
||
|
conv = scipy.signal.fftconvolve(convol_data, pattern_data)
|
||
|
if len(conv) == 0:
|
||
|
delete_list.append(segment_index)
|
||
|
continue
|
||
|
upper_bound = self.state.convolve_max * (1 + POSITIVE_SEGMENT_MEASUREMENT_ERROR)
|
||
|
lower_bound = self.state.convolve_min * (1 - POSITIVE_SEGMENT_MEASUREMENT_ERROR)
|
||
|
delete_up_bound = self.state.conv_del_max * (1 + NEGATIVE_SEGMENT_MEASUREMENT_ERROR)
|
||
|
delete_low_bound = self.state.conv_del_min * (1 - NEGATIVE_SEGMENT_MEASUREMENT_ERROR)
|
||
|
max_conv = max(conv)
|
||
|
if max_conv > upper_bound or max_conv < lower_bound:
|
||
|
delete_list.append(segment_index)
|
||
|
elif max_conv < delete_up_bound and max_conv > delete_low_bound:
|
||
|
delete_list.append(segment_index)
|
||
|
|
||
|
for item in delete_list:
|
||
|
segments_indexes.remove(item)
|
||
|
segments_indexes = utils.remove_duplicates_and_sort(segments_indexes)
|
||
|
return segments_indexes
|