Skip to main content

FireHole (fh) is a Python package for Data Analyst.

Project description

FireHole :fire::hole:

中文 | docker

FireHole is a Python package customized for data analysts.

Installation

pip install firehole

FireHole requires Python 3.4 or later

Quick Start

Introducing FireHole

import firehole as fh
# FireHole is based on numpy, so we also import it
import numpy as np

Scoring of Comprehensive System

Evaluate and score systems according to the following methods:

  • Analytic Hierarchy Process (AHP)
  • Entropy Weight Method
  • Coefficient of Variation Method
  • ......

1. AHP

import json
import firehole as fh

model = {
    "name": "Sample Model",
    "method": "approximate",
    "criteria": ["critA", "critB", "critC"],
    "subCriteria": {
        "critA": ["subCritA", "subCritB"]
    },
    "alternatives": ["altA", "altB", "altC"],
    "preferenceMatrices": {
        "criteria": [
            [1, 1, 1],
            [1, 1, 1],
            [1, 1, 1]
        ],
        "subCriteria:critA": [
            [1, 1],
            [1, 1]
        ],
        "alternatives:subCritA": [
            [1, 1, 1],
            [1, 1, 1],
            [1, 1, 1]
        ],
        "alternatives:subCritB": [
            [1, 1, 1],
            [1, 1, 1],
            [1, 1, 1]
        ],
        "alternatives:critB": [
            [1, 1, 1],
            [1, 1, 1],
            [1, 1, 1]
        ],
        "alternatives:critC": [
            [1, 1, 1],
            [1, 1, 1],
            [1, 1, 1]
        ]
    }
}

ahp_model = fh.parse(model)
priorities = ahp_model.get_priorities()

2. Entropy Weight Method

>>> li = [[100, 90, 100, 84, 90, 100, 100, 100, 100],
          [100, 100, 78.6, 100, 90, 100, 100, 100, 100],
          [75, 100, 85.7, 100, 90, 100, 100, 100, 100],
          [100, 100, 78.6, 100, 90, 100, 94.4, 100, 100],
          [100, 90, 100, 100, 100, 90, 100, 100, 80],
          [100, 100, 100, 100, 90, 100, 100, 85.7, 100],
          [100, 100, 78.6, 100, 90, 100, 55.6, 100, 100],
          [87.5, 100, 85.7, 100, 100, 100, 100, 100, 100],
          [100, 100, 92.9, 100, 80, 100, 100, 100, 100],
          [100, 90, 100, 100, 100, 100, 100, 100, 100],
          [100, 100, 92.9, 100, 90, 100, 100, 100, 100]]
>>> # convert list into numpy array
>>> li = np.array(li)
>>> # calculation by entropy method
>>> entropy = fh.Entropy()
>>> rlt = entropy.entropy(li)
>>> 
>>> print(rlt['entropy'])
[0.9912947 0.99752157 0.98629816 0.99731237 0.9946718 0.99902487 0.95993528 0.9978992 0.99556829]
>>> print(rlt['weight'])
[0.10817562 0.03079802 0.17026463 0.03339765 0.06621039 0.0121174 0.49786068 0.02610541 0.0550702 ]
>>> print(rlt['score'])
[1.09856723 0.31796876 1.4625885 0.35533899 0.59410402 0.1305623 4.91581354 0.27875173 0.58096578]

3. Coefficient of Variation Mehtod

>>> a = [[3051, 20.4, 30.1, 0.5250, 31.1, 9.5, 70.1, 82.8, 6, 2],
         [20763, 0.1, 84.1, 0.9969, 100, -1.7, 78.6, 92.9, 28, 1.3],
         [2407, 16.7, 39.9, 0.5504, 38.8, 15.6, 65.6, 85.7, 7, 0.2],
         [5121, 24.9, 38.4, np.nan, 60.6, 16.5, 69.5, 74.6, 18, 0.9],
         [16861, np.nan, np.nan, 0.9769, 91, 15.7, 77.9, 95.7, 44, 4.6],
         [23592, 1.9, 60.3, 0.9473, 78.5, 2.1, 80, 99, 43, 1.8],
         [4317, 12.8, 60.3, np.nan, 56.4, 4.4, 67.9, 99, 32, 3.5],
         [13286, 5.8, 51.2, 0.8824, 80.4, 8.7, 72.6, 97.5, 60, 1.1],
         [7699, 12.6, 42, 0.8120, 55.9, 20.5, 72.2, 86.4, 11, 0.5],
         [3725, 20.6, 47.3, 0.6012, 56.8, 22.3, 68.6, 94.8, 35, 0.6],
         [25295, 0.2, 65.2, 0.9979, 100, 8.6, 77.3, 91.8, 39, 1.7],
         [2945, 22.4, 51.1, 0.5843, 23, 11.5, 73.3, 91.1, 5, 4.2],
         [5524, 9.6, 49.5, 0.4875, 21, 10.2, 68.9, 95, 21, 2.9],
         [6594, 17.4, 54.7, 0.5658, 72.9, 14.8, 69.3, 84, 18, 1.5],
         [3146, 17.3, 51.1, np.nan, 44.9, 17.2, 66.7, 53.7, np.nan, 2.1],
         [8296, 4.2, 62.2, np.nan, 52.9, 15.9, 53.2, 84.6, 19, 0.6],
         [22814, np.nan, np.nan, 0.9651, 76.9, 4.4, 79.1, 99, 90, 2.1],
         [7450, 5.5, 68.6, 0.8021, 74, 22.4, 72.3, 90.8, 16, 1.2],
         [29240, 1.8, 71.8, 0.9733, 76.8, 5.8, 76.8, 99, 81, 2.6],
         [11728, 6, 65.6, 0.9849, 89.3, 11.7, 73.1, 96.7, 42, 2.7],
         [6460, 8.3, 62.3, 0.7581, 80.1, 13.0, 67, 84.5, 12, 1.3],
         [5706, 4.2, 49.3, 0.8921, 86.3, 20.5, 72.6, 92, 25, 2.4],
         [6314, 16, 44, np.nan, 70.6, -4.3, 68.1, 99.5, 44, 4.3],
         [4683, 14.5, 57, np.nan, 69, -6.4, 71.3, 98.2, 41, 3.5],
         [12197, 4.3, 58, 0.9535, 74.6, -1.8, 74.1, 99, 23, 2.9],
         [21214, 2.3, 71.5, np.nan, 75.2, 3.4, 78.2, 99, 52, 2.9],
         [22026, 1.1, 44.2, 0.9717, 87.1, -0.7, 77.3, 99, 45, 3.4],
         [20365, 2.8, 66.4, np.nan, 66.8, -0.1, 78.3, 98.3, 43, 5.5],
         [22325, np.nan, np.nan, 0.9681, 89.2, 3.1, 78, 99, 50, 2.6],
         [7543, 5.6, 61.6, 0.8089, 64.8, 0.6, 72.7, 99.7, 24, 2.3],
         [5572, 19.5, 36.1, 0.6003, 55.7, -1.3, 70.2, 97.9, 23, 1.8],
         [6180, 7.5, 55.7, np.nan, 77, -4.7, 66.7, 99.5, 41, 4.6],
         [15960, 3.5, 25.1, 0.9239, 77.2, 0.1, 78.1, 97.4, 51, 4.2],
         [20314, 1.8, 66.7, 0.9835, 89.4, 1.3, 77.3, 99, 50, 1.6],
         [3130, 13.8, 48.4, 0.7388, 67.8, -6.3, 69.1, 99.6, 42, 4.5],
         [21795, 3.2, 70.6, 0.9524, 84.7, 6.5, 78.3, 99, 76, 2.5],
         [16084, np.nan, np.nan, 0.9148, 85.6, 7.9, 77.1, 99, 59, 2.1]]
>>> a = np.array(a)
>>> target = np.array([22930, 2, 63, 0.9580, 76, 4, 77, 97.5, 58, 2.5], dtype='float')
>>> pos_neg = np.array([1, -1, 1, 1, 1, -1, 1, 1, 1, 1], dtype='float')
>>> cov = fh.COV()
>>> result = cov.comprehensive(data=a, target=target, pos_neg=pos_neg, decimals=2)
>>> print(result)
[ 0.35  3.3   0.25  0.3   0.63  1.18  0.64  0.62  0.32  0.33  2.36  0.45
  0.46  0.4   0.3   0.39  0.81  0.43  1.04  0.64  0.43  0.52  0.25  0.28
 -0.03  0.98 -0.54 -9.31  0.85  2.12 -0.45  0.31 10.72  1.48  0.35  0.88
  0.6 ]

Keyword Extraction

Extract keywords from text based on Trie dicitonary data structure.

1. Extract keywords

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_processor.add_keyword('Big Apple', 'New York')
>>> keyword_processor.add_keyword('Bay Area')
>>> keywords_found = keyword_processor.extract_keywords('I love Big Apple and Bay Area.')
>>> keywords_found
>>> # ['New York', 'Bay Area']
import firehole as fh
keyword_processor = fh.KeywordProcessor()
keyword_processor.add_keyword('口号')
keyword_processor.add_keyword('横幅')
keywords_found = keyword_processor.extract_keywords('他边喊口号边走近了那个建筑,并拉起了横幅。')
keywords_found
# ['口号', '横幅']

2. Replace keywords

>>> keyword_processor.add_keyword('New Delhi', 'NCR region')
>>> new_sentence = keyword_processor.replace_keywords('I love Big Apple and new delhi.')
>>> new_sentence
>>> # 'I love New York and NCR region.'

3. Case Sensitive example

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor(case_sensitive=True)
>>> keyword_processor.add_keyword('Big Apple', 'New York')
>>> keyword_processor.add_keyword('Bay Area')
>>> keywords_found = keyword_processor.extract_keywords('I love big Apple and Bay Area.')
>>> keywords_found
>>> # ['Bay Area']

4. Span of keywords extracted

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_processor.add_keyword('Big Apple', 'New York')
>>> keyword_processor.add_keyword('Bay Area')
>>> keywords_found = keyword_processor.extract_keywords('I love big Apple and Bay Area.', span_info=True)
>>> keywords_found
>>> # [('New York', 7, 16), ('Bay Area', 21, 29)]

5. Get Extra information with keywords extracted

>>> import firehole as fh
>>> kp = fh.KeywordProcessor()
>>> kp.add_keyword('Taj Mahal', ('Monument', 'Taj Mahal'))
>>> kp.add_keyword('Delhi', ('Location', 'Delhi'))
>>> kp.extract_keywords('Taj Mahal is in Delhi.')
>>> # [('Monument', 'Taj Mahal'), ('Location', 'Delhi')]
>>> # NOTE: replace_keywords feature won't work with this.

6. No clean name for Keywords

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_processor.add_keyword('Big Apple')
>>> keyword_processor.add_keyword('Bay Area')
>>> keywords_found = keyword_processor.extract_keywords('I love big Apple and Bay Area.')
>>> keywords_found
>>> # ['Big Apple', 'Bay Area']

7. Add Multiple Keywords simultaneously

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_dict = {
>>>     "java": ["java_2e", "java programing"],
>>>     "product management": ["PM", "product manager"]
>>> }
>>> # {'clean_name': ['list of unclean names']}
>>> keyword_processor.add_keywords_from_dict(keyword_dict)
>>> # Or add keywords from a list:
>>> keyword_processor.add_keywords_from_list(["java", "python"])
>>> keyword_processor.extract_keywords('I am a product manager for a java_2e platform')
>>> # output ['product management', 'java']

8. To Remove keywords

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_dict = {
>>>     "java": ["java_2e", "java programing"],
>>>     "product management": ["PM", "product manager"]
>>> }
>>> keyword_processor.add_keywords_from_dict(keyword_dict)
>>> print(keyword_processor.extract_keywords('I am a product manager for a java_2e platform'))
>>> # output ['product management', 'java']
>>> keyword_processor.remove_keyword('java_2e')
>>> # you can also remove keywords from a list/ dictionary
>>> keyword_processor.remove_keywords_from_dict({"product management": ["PM"]})
>>> keyword_processor.remove_keywords_from_list(["java programing"])
>>> keyword_processor.extract_keywords('I am a product manager for a java_2e platform')
>>> # output ['product management']

9. To check Number of terms in KeywordProcessor

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_dict = {
>>>     "java": ["java_2e", "java programing"],
>>>     "product management": ["PM", "product manager"]
>>> }
>>> keyword_processor.add_keywords_from_dict(keyword_dict)
>>> print(len(keyword_processor))
>>> # output 4

10. To check if term is present in KeywordProcessor

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_processor.add_keyword('j2ee', 'Java')
>>> 'j2ee' in keyword_processor
>>> # output: True
>>> keyword_processor.get_keyword('j2ee')
>>> # output: Java
>>> keyword_processor['colour'] = 'color'
>>> keyword_processor['colour']
>>> # output: color

11. Get all keywords in dictionary

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_processor.add_keyword('j2ee', 'Java')
>>> keyword_processor.add_keyword('colour', 'color')
>>> keyword_processor.get_all_keywords()
>>> # output: {'colour': 'color', 'j2ee': 'Java'}

12. To set or add characters as part of word characters

>>> import firehole as fh
>>> keyword_processor = fh.KeywordProcessor()
>>> keyword_processor.add_keyword('Big Apple')
>>> print(keyword_processor.extract_keywords('I love Big Apple/Bay Area.'))
>>> # ['Big Apple']
>>> keyword_processor.add_non_word_boundary('/')
>>> print(keyword_processor.extract_keywords('I love Big Apple/Bay Area.'))
>>> # []

Text Similarity

Calculate the similarity between different sentence.

1. BM25

  • OKapi BM25
  • BM25L
  • BM25+
>>> import firehole as fh
>>> corpus = ["Hello there good man!", 
              "It is quite windy in London", 
              "How is the weather today?"]
>>> tokenized_corpus = [doc.split(" ") for doc in corpus]
>>> bm25 = fh.BM25_okapi(tokenized_corpus)
<firehole.algorithms.similarity.rank_bm25.BM25Okapi at 0x13a49c88>
>>> query = "windy London"
>>> tokenized_query = query.split(" ")
>>> doc_scores = bm25.get_scores(tokenized_query)
# array([0.    , 0.93729472, 0.    ])
>>> bm25.get_top_n(tokenized_query, corpus, n=1)
# ['It is quite windy in London']
import firehole as fh
import jieba
corpus = ["他手执横幅走了进来", 
          "他走了进来", 
          "横幅上写着很多的字"]
tokenized_corpus = [jieba.lcut(doc, cut_all=False, HMM=True) for doc in corpus]
bm25 = fh.BM25Okapi(tokenized_corpus)
query = "他有一个横幅"
tokenized_query = jieba.lcut(query, cut_all=False, HMM=True)
doc_scores = bm25.get_scores(tokenized_query)
print(doc_scores)
# [0.04147104 0.02453117 0.01924653]
bm25.get_top_n(tokenized_query, corpus, n=1)
# ['他手执横幅走了进来']

2. Simhash

View simhash value:

import re
import firehole as fh

def get_features(s):
    width = 3
    s = s.lower()
    s = re.sub(r'[^\w]+', '', s)
    return [s[i:i + width] for i in range(max(len(s) - width + 1, 1))]

text1 = 'How are you? I am fine. Thanks.'
text2 = 'How are u, I am fine. Thanks.'
text3 = 'How r u?I am fine. Thanks.'

print('%x' % fh.Simhash(get_features(text1)).value)
print('%x' % fh.Simhash(get_features(text2)).value)
print('%x' % fh.Simhash(get_features(text3)).value)
import re
import firehole as fh

def get_features(s):
    width = 3
    s = re.sub(r'[^\w]+', '', s)
    return [s[i:i + width] for i in range(max(len(s) - width + 1, 1))]

text1 = '他手执横幅走了进来。'
text2 = '他走了进来。'
text3 = '横幅上写着很多的字'

print('%x' % fh.Simhash(get_features(text1)).value)
print('%x' % fh.Simhash(get_features(text2)).value)
print('%x' % fh.Simhash(get_features(text3)).value)
# f8b6c362c7552669
# fa377b62e7d9a81c
# 98965516240442ee

Get distance of two simhash:

import re
import firehole as fh

print(fh.Simhash('aa').distance(fh.Simhash('bb')))
print(fh.Simhash('aa').distance(fh.Simhash('aa')))

hash1 = fh.Simhash(u'I am very happy'.split())
hash2 = fh.Simhash(u'I am very sad'.split())
print(hash1.distance(hash2))
import re
import firehole as fh

print(fh.Simhash('政府').distance(fh.Simhash('大楼')))
print(fh.Simhash('政府').distance(fh.Simhash('政府')))

hash1 = fh.Simhash(jieba.lcut("我很开心", cut_all=False, HMM=True))
hash2 = fh.Simhash(jieba.lcut("我很忧伤", cut_all=False, HMM=True))
print(hash1.distance(hash2))

Use the SimhashIndex to query near duplicates objects in a very efficient way:

import re
import firehole as fh

def get_features(s):
    width = 3
    s = s.lower()
    s = re.sub(r'[^\w]+', '', s)
    return [s[i:i + width] for i in range(max(len(s) - width + 1, 1))]

data = {
    1: u'How are you? I Am fine. blar blar blar blar blar Thanks.',
    2: u'How are you i am fine. blar blar blar blar blar than',
    3: u'This is simhash test.',
}
objs = [(str(k), fh.Simhash(get_features(v))) for k, v in data.items()]
index = fh.SimhashIndex(objs, k=3)

print(index.bucket_size())

s1 = fh.Simhash(get_features(u"How are you i am fine. blar blar blar blar blar thank"))
print(index.get_near_dups(s1))

index.add("4", s1)
print(index.get_near_dups(s1))

Format Converter

1. ID converter (between 15 & 18 digits)

Convert single ID number between 15 digits and 18 digits format:

import firehoe as fh
converter = fh.ConvertID()
# ID number of 18 digits
test_id1 = "130701199310302288"
# ID number of 15 digits
test_id2 = "320311770706001"
# convert from 15 digits to 18 digits
new_18 = converter.up_to_eighteen(test_id2)
print("Convered to 18 digits: ", new_18)
# convert from 18 digits to 15 digits
new_15 = converter.down_to_fifteen(test_id1)
print("Convered to 18 digits: ", new_18)

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

firehole-0.4.2.tar.gz (28.8 kB view details)

Uploaded Source

Built Distribution

firehole-0.4.2-py3-none-any.whl (34.0 kB view details)

Uploaded Python 3

File details

Details for the file firehole-0.4.2.tar.gz.

File metadata

  • Download URL: firehole-0.4.2.tar.gz
  • Upload date:
  • Size: 28.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.5.0.1 requests/2.22.0 setuptools/41.4.0 requests-toolbelt/0.9.1 tqdm/4.36.1 CPython/3.7.4

File hashes

Hashes for firehole-0.4.2.tar.gz
Algorithm Hash digest
SHA256 331f9aae7b3b162b90e3d824492994bca498472cc7876e38885b1dbad13b0429
MD5 8614a51610d8d6fe7a2b66a8541487e0
BLAKE2b-256 d502f7871468b135a8a1638065c071ee671c573228c7f84e5113c4600c5476f0

See more details on using hashes here.

File details

Details for the file firehole-0.4.2-py3-none-any.whl.

File metadata

  • Download URL: firehole-0.4.2-py3-none-any.whl
  • Upload date:
  • Size: 34.0 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/3.1.1 pkginfo/1.5.0.1 requests/2.22.0 setuptools/41.4.0 requests-toolbelt/0.9.1 tqdm/4.36.1 CPython/3.7.4

File hashes

Hashes for firehole-0.4.2-py3-none-any.whl
Algorithm Hash digest
SHA256 55ecd9b23f1d907faafa9693ff595a83bab8cb88f50d08cae028c57538c2b5df
MD5 81eb99ebd87ca777d63e249bcfe84112
BLAKE2b-256 3349df6fa2b1a98a0d4eb7fe3e239e0b42219e80646b730d48f36d7fe63dd25d

See more details on using hashes here.

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page