Скачать или смотреть What is Tfidf Matrix and TfidfVectorizer

TF- IDF is a representation of the text documents in the form of vectors. It is dependent on the significance of the word. Those vectors form a matrix with the most significant and less significant words in the document. It is nothing but a formula based on Statistics.

TF-IDF is applied in Natural language processing, Machine learning tasks to assist those algoriths in predicting the output.

TF-IDF Statistical formula is split into two portions. one with the Term Frequency(TF) and another with the Inverse Document Frequency (IDF)

Term Frequency(TF)
It is the ratio of the total no of occurences of the word (wo) in the document to the total no of words in the document

IDF: It is measured by dividing the total no of documents in the dataset/corpus by the total no of documents containing the word in it.
i.e. ln(total no of documents in the dataset(corpus) / no of documents containing the word (wo))

So TF-IDF is the product of the Term Frequency and the inverse document frequency.