pattern_search

 1from turtle import distance, title
 2from unittest import result
 3import pandas as pd
 4import pattern_utils
 5import Pattern as p
 6import tendencyCalculator as tc
 7import matplotlib.pyplot as plt
 8from random import randint
 9import normalize_utils as nu
10
11INCREMENT = 1
12    
13def findCurrentPatterns(start_index, finish_index, company_dataframe, patterns_dictionary, company_name, intensive_search):
14    """Find patterns that are occuring the day of the search  
15    
16    Args:  
17        start_index (int): starting size of the searching window
18        finish_index (int): ending size of the searching window
19        company_datadrame (dataframe): dataframe containing the company prices in the entire year
20        patterns_dictionary (Dict[]): dictionary containing types of patterns as keys an pattern data as value
21        company_name (str): name of the company where the search is taking place  
22    Returns:  
23        results (List[Pattern]): list of patterns found
24    """
25    distances_dict = {
26        'double_top': [pattern_utils.BIG_NUMBER, -1],
27        'double_bottom': [pattern_utils.BIG_NUMBER, -1]
28    }
29    patterns_dictionary_copy = patterns_dictionary.copy()
30    results = []
31    for i in range(finish_index, start_index, -1): # hay que añadir que itere por todos y que se quede con el de menor distancia 
32        if len(patterns_dictionary_copy.keys()) == 1:
33            break
34        if i > company_dataframe.size:
35            i = company_dataframe.size
36        sliced_dataframe = company_dataframe.iloc[company_dataframe.size - i:]
37        normalized_vector = nu.normalizeVector(sliced_dataframe['Close'].tolist())
38        new_pattern_type, distance = pattern_utils.findCommonPattern(normalized_vector, patterns_dictionary_copy)
39        if new_pattern_type != 'rest_normalized':
40            if intensive_search == True:
41                if distance < distances_dict[new_pattern_type][0]:
42                    distances_dict[new_pattern_type] = [distance, i]
43            else:
44                distances_dict[new_pattern_type] = [distance, i]
45                patterns_dictionary_copy.pop(new_pattern_type)
46
47    for key in distances_dict:
48        if distances_dict[key][1] == -1:
49            continue
50        new_pattern = p.Pattern(key, company_dataframe[company_dataframe.size - distances_dict[key][1]:], company_name, str(company_dataframe.iloc[company_dataframe.size - i].name), str(company_dataframe.iloc[company_dataframe.size - 1].name), None)
51        results.append(new_pattern)
52            #patterns_dictionary_copy.pop(new_pattern_type)
53
54    return results
55
56def findHistoricPatterns(window_width, company_data, patterns_dictionary, company_name):
57    """Find patterns through historic data  
58    Args:  
59        window_width (int): fixed window size for the search
60        company_data (datarame): dataframe containing the company's close prices
61        atterns_dictionary (Dict[]): dictionary containing types of patterns as keys an pattern data as value
62        company_name (str): name of the company where the search is taking place  
63    Returns:  
64        patterns_found (List[Pattern]): list of patterns found
65    """
66    patterns_found = []
67    offset = 0
68    i = 0
69    while i < company_data.size - window_width - 1:
70        right_window_index = i + window_width
71        if right_window_index >= company_data.size:
72            break
73        sliced_dataframe = company_data.iloc[i:right_window_index]
74        normalized_vector = nu.normalizeVector(sliced_dataframe['Close'].tolist())
75        new_pattern_type, best_distance_found = pattern_utils.findCommonPattern(normalized_vector, patterns_dictionary)
76        if new_pattern_type != 'rest_normalized' and new_pattern_type != '':
77            left_index, right_index = pattern_utils.enhanceDataframe(best_distance_found, new_pattern_type, sliced_dataframe['Close'].tolist(), patterns_dictionary, [1,2,3,4])
78            dataframe_segment = sliced_dataframe[left_index:right_index] #Esto sin ventana mejorada
79            longer_dataframe = company_data[i + left_index:] #Quitar left_index si no se usa enhanced dataframe
80            pattern_tendency = tc.findPatternTendency(dataframe_segment, longer_dataframe, new_pattern_type)
81            if pattern_tendency != None:
82                new_pattern = p.Pattern(new_pattern_type, pattern_tendency[1], company_name, str(dataframe_segment.iloc[0].name), str(dataframe_segment.iloc[dataframe_segment.size - 1].name), pattern_tendency[0])
83                patterns_found.append(new_pattern)
84            offset = sliced_dataframe.size
85        else:
86            offset = 0
87        i += INCREMENT + offset
88    return patterns_found