Natural Language Processing|TF-IDF for Machine Learning| Text Prerocessing

[@neharikasrivastav7465]

your channel deserves a lot of love. seriously one of the best explanations of NLP.

[@akashpoudel571]

Very lucid explaination sir....u are the best of all instructor....any body would understand Data Science having u as a guide... So blessed to watch your explaination

[@ArchnaVijay]

Honestly I searched many videos to understand TFIDF .. this video is “The best” among the rest !

[@premranjan4440]

Sir, I love your work. I am currently doing specilization in Data Science and AI and I am learned more from than my two years of college. Keep up the good work sir!

[@mathketeer]

Thank you. I love the fact that you do 2 videos: concept explanation and programming explanation.

[@srinivaspachika1996]

Just a conclusion saying rare words will have high tf-idf scores will be a good addition. Nonetheless, This is the best video I have seen so far on Tf-Idf . thank you for such amazing content

[@arpittiwari6590]

Hey Krish, You are the best instructor I had ever seen, You deserve more and more. You explain each thing in a way that it should be. I gonna will be one of the members all the times, I have learnt something in each of your videos.

[@arjyabasu1311]

Thanks a lot sir... your contribution to the data science world is just priceless !!!

[@bijaynayak6473]

Explained very clearly and Nicely Thanks a TON Krish Naik

[@dnakhawa]

You are the Best sir in Data Science

[@samrat_chauhan]

Sir your core knowledge, of what an student is expecting is very good

[@ishantyadav5532]

Sir this is the best playlist to learn about NLP and saves a lot of time into researching source material. I am referring to it for my internship. Really underrated channel; I wish you much more subscribers and support.

[@rishieee887]

Best NLP course, even for post graduate course

[@shahidmalik6107]

thank you so much for a brief and easy lecture. hats off from Pakistan

[@girishreddyedula2667]

You are the real savior Krish Never ever seen this beauty in explaining NLP hope to see more applications in NLP in the near future

[@arijitRC473]

Sir, really this kind of videos help us a lot.....really you are doing a great job, your videos insist data science enthusiasts a lot to become a data scientist

[@sindhunannapaneni5715]

Thanq so much to take a step forward to help all who thrive to learn data science, kudos to your great efforts, your videos are really helpful and am gaining knowledge as well as confidence by watching your videos..God bless you Krish

[@daneshrepalle6408]

Really your explanation is Great.In sentiment analysis, Expression is very important might be when you converted the data into lower case, shouted expression will lose.

[@akshaytakhi8016]

Beautifully explained, best data science mentor of all time!!!! Sir please make a video for amazon review sentiment analysis. That would be really helpfull for us. Thanks

[@pankushkukreja3101]

Thank you krish, it very helpfull. Request if you can add below topics in NLP Playlist: Latent Dirichlet Algorithm (LDA) Latent Semantic Analysis (LSA) Singular Value Decomposition (SVD) Word Embeddings (Word2vec and GloVe)

[@karthikplrao2715]

It is very better much better than udemy

[@karthip23]

Amazing Content. No words to thank for explaining so beautifully :)

[@kayodeoyedele1594]

This is great..You are the best man ..Really nice videos

[@sriharshavardhan299]

the best videos for ML

[@sandipansarkar9211]

Superb video Krish to contribute to understand of NLP.Thanks

[@injetiprasad8937]

Really Awesome, good explanation. Thank you so much. Actually, u said , in IDF part "be" containing in 4 times but u explained in 3 times only.

[@karthikplrao2715]

Please make a vedio on pca and tda

[@rajsinghmaan3095]

In this code 'word' used in WordNetLemmatizer is not defined. It is used in 3rd tutorial for word tokenization, but if a new person runs code from the above video, it leads to an error.

[@AkshatSingh0501]

Here one more thing is to add which Krish didn't add is CountVectorizer will be added before Tfidf vectorizer or else it will give error of "Vocabulary empty ".

[@anantchourasia3802]

Bro that was an awesome explanation 🤐 Keep it up ✌🏻

[@sudhanshuedu]

Great explanation

[@sandipansarkar9211]

Superb video for beginners

[@thiagoribeiro4733]

Great video Krish, thanks for your effort!

[@rahul4upandey]

please share some more recent model in NLP like BERT, Transformer

[@karthikplrao2715]

It is very nice

[@dharmendrasingh-iz1by]

Excellent explanation!

[@flaviadecarvalhoneves7541]

Awesome video. Congrats!! Very helful!

[@krishj8011]

Great Tutorial...

[@adityasharma2667]

Fantastic Sir, you are making Data Science easier and easier day by day. One query Sir. when we convert the corpus using Tf-IdfVecorizer or Counter Vecorizer we got array of shape (31,114) what is this 114?

[@ashishjadhav4028]

114 is the number of columns

[@anandtalware2283]

Hello, good evening, should we apply PCA after vectorization(BOW or TFIDF) ? which accuracy is better , with or without PCA ?

[@HenokGashaw]

Excellent explanation!!!!!

[@maulindusarkar4581]

I used TF-IDF on the para you considered here, but got something like this: paragraph = """The boy is good. The girl is good. The boy and girl are good""" array([[0.78980693, 0. , 0.61335554], [0. , 0.78980693, 0.61335554], [0.61980538, 0.61980538, 0.48133417]]) But I should get two zeroes in the first and second rows right, according to the formula?

[@sagarghimire1174]

Best explanation. Thanks

[@sudeshnadutta5702]

Hi Krish. I had one question. There's a parameter called max_features inside CountVectorizer as well as tdidfvectorizer. How does that work? I am assuming that when the frequency distribution is calculated and sorted then we can choose the top 'n' features? Is that correct? Please let me know. Thank you

[@gulsanbor]

You are excellent

[@akashpoudel571]

Waiting for ua upload on sentiment analysis from amazon Sir....

[@shubhamteke2856]

Sir i got 1 error like 'list' object has no attribute 'lower'. How to solve this error

[@salmanhaider962]

I m getting error while importing TfidVectorizer package.

[@bradleyadjileye1202]

Big thanks for this one

[@arnoldnana7076]

Great video

[@amruthasankar3453]

Thankyou sir❤️🔥

[@souvikdas7200]

Great explanation sir. I've read a few papers where this tf-idf matrix is used to do the text clustering by means of Kmeans clustering . I'm just wondering how it's actually working out. Would you please explain it in a video?

[@krishnaik06]

After u get the vectors apply K means clustering

[@souvikdas7200]

@@krishnaik06 Thank you very much sir. I'm trying this. Will it be okay if I find cosine similarity matrix from the tf-idf and then apply Kmeans clustering?

[@krishnaik06]

Yes go ahead...u can try anything

[@kuldeeppatle8731]

Sir please make video on RAKE?

[@ga43ga54]

Can you please make a video on Word2Vec.... Thank you !!

[@suvarnadeore8810]

Thank you krish sir

[@rbattula417]

Thanks for sharing knowledge. Could you please share more content for NLP and if possible Deep Learning for OCR.

[@krishnaik06]

Hi Rajshekar , I am planning to upload many videos. Stay tuned.

[@bigj3867]

@@krishnaik06 thank you sir..

[@manojjena1903]

is the code workws for odia languagae text

[@ayushsingh-qn8sb]

If we have lemmatizer why do we even need stemming

[@dasarithejaswaroop1072]

U R SUPER SIR

[@muhammadusmanakram406]

sir link to sentimental analysis ection???

[@preetyk7615]

thanks ,very informative vedio sir,here i have one doubt in IDF we are doing log in which the frequency is in denominator that means as freq of the word in doc is more the idf value will be less that means indirectly proportional that means idf*tf will go down which mean for more freq word the idf*tf value is less then one whose freq is less,that is not correct right ??? for more freq word used in doc its idf*tf product shoul dbe high.

[@GokulThiagarajan1]

Hi Preety, your question is correct. This model gives more priority to unique words than common words. Hence works different than bag of words and unique words will have more score sometimes than common words

[@smvignesh3650]

Great Video!!!

[@alphonseinbaraj7602]

Sir ,Your Data Preprocessing Techniques playlist was removed .why ? kindly revert it . I am learner .request you to revert that . Because without knowing DATA PREPROCESSING and without understanding about it ,shouldnt go furthr .please kindly revert it.

[@datascience3008]

Thanks krish

[@bismeetsingh352]

I understand tf idf but still the intuition as to when and how to use the information for further analysis is not clear. COULD YOU KINDLY EXPLAIN THAT?

[@DS_AIML]

As per my understanding,TF*IDF will give more weightage to words which is more important then other words.For example if i want to know what is first sentence in vedio talking about,i will come to know that 'boy' is given more emphasis then word 'good'

[@funtime12345]

Thank you sir!

[@dhirajsharma74]

Hi, Thank you so much for the awesome video. I am getting mention below error, could you please help me with it. Thanks Error : ""

[@dhirajsharma74]

Error: Expected 2D array, got 1D array instead

[@shivamsrivastava416]

Sir , could you please share your PPT

[@debatradas9268]

thanks

[@indirajithkv7793]

❤

[@david76383]

It does not work for me? The final vector still shows only valus of 1 and 0

[@ghali3059]

The sound is terrible its hurting my ears.

[@Anjy2709]

so u haven't implemented the tf-idf completely..i could not understand where is vocabulary

[@himsinghvi88]

Hi @Krish I have tried using the tfidf on the boy-girl example, but it is not doing 0 for the word "good", I am getting following result. ['boy', 'girl', 'good'] array([[0.78980693, 0. , 0.61335554], [0. , 0.78980693, 0.61335554], [0.61980538, 0.61980538, 0.48133417]]) why it is so ?