You will be an engineer in 100 days --Day 68 --Programming --About TF-IDF

Click here until yesterday

This time about TF-IDF.

What is TF-IDF?

TF-IDF is the index term frequency and the reverse document frequency. TF (Term Frequency) word frequency and ʻIDF (Inverse Document Frequency)` The rarity of words It will be a product.

{TF: Frequency of occurrence of specified words in document: \ frac {Number of occurrences of specified words in document} {Number of occurrences of all words in document} \\ }

{IDF: Inverse document frequency (rareness of specified word): log \ frac {total number of documents} {number of documents including specified word}}

{TFIDF (index term frequency reverse document frequency) = TF * IDF}

reference: https://ja.wikipedia.org/wiki/Tf-idf

Word count

First, let's count the number of words in the sentence.

Make a sentence for counting.

result_list = []
result_list.append('I am a cat')
result_list.append('I am a cat')
result_list.append('I am also')
result_list.append('Please, please be a cat')

You can count the frequency of occurrence of words with the following code.

from sklearn.feature_extraction.text import CountVectorizer
import numpy as np
import pandas as pd
import warnings
warnings.filterwarnings('ignore')

count_vectorizer = CountVectorizer(token_pattern='(?u)\\b\\w+\\b')

count_vectorizer.fit(result_list)
X = count_vectorizer.transform(result_list)

print(len(count_vectorizer.vocabulary_))
print(count_vectorizer.vocabulary_)

pd.DataFrame(X.toarray(), columns=count_vectorizer.get_feature_names())

8 {'I'm: 6,'is': 4,'cat': 7,'in': 1,'is': 0,'also': 5,'is': 2,'please': 3}

	is there	so	is	here you go	Is	Also	I	Cat
0	1	1	0	0	1	0	1	1
1	1	1	0	0	1	0	1	1
2	0	0	1	0	0	1	1	0
3	0	0	1	2	0	0	0	1

You can count how many times a word appears in each sentence.

Next, let's find TF-IDF.

You can find it with the following code.

from sklearn.feature_extraction.text import TfidfVectorizer
import warnings
import numpy as np
import pandas as pd
warnings.filterwarnings('ignore')

tfidf_vectorizer = TfidfVectorizer(token_pattern='(?u)\\b\\w+\\b')
tfidf_vectorizer.fit(result_list)

print(len(tfidf_vectorizer.vocabulary_))
print(tfidf_vectorizer.vocabulary_)

X = tfidf_vectorizer.fit_transform(result_list)
pd.DataFrame(X.toarray(), columns=tfidf_vectorizer.get_feature_names())

8 {'I'm: 6,'is': 4,'cat': 7,'in': 1,'is': 0,'also': 5,'is': 2,'please': 3}

	is there	so	is	here you go	Is	Also	I	Cat
0	0.481635	0.481635	0	0	0.481635	0	0.389925	0.389925
1	0.481635	0.481635	0	0	0.481635	0	0.389925	0.389925
2	0	0	0.553492	0	0	0.702035	0.4481	0
3	0	0	0.35157	0.891844	0	0	0	0.284626

TF-IDF has a value between 0 and 1. The value of what appears in many sentences is small. What appears a lot in one sentence is considered to be an important word.

You can think of it as a word that is more rare as it gets closer to 1.

Summary

There are methods for vectorizing sentences and methods for calculating the rarity of words. Since the sentences can be quantified, you will be able to perform various calculations.

Because these methods are often used in machine learning etc. Let's suppress the name and so on.

32 days until you become an engineer

Author information

Otsu py's HP: http://www.otupy.net/

Youtube： https://www.youtube.com/channel/UCaT7xpeq8n1G_HcJKKSOXMw

Twitter： https://twitter.com/otupython