Тёмный
Devvrat Rana
Devvrat Rana
Devvrat Rana
Подписаться
I am a Data Science Researcher having more than 10 year experience in IT industry.
I have worked on several classical, rule base and AI projects by using Machine Learning, Natural Language Processing, Transformers and Deep Learning, Spark, Kafka, AWS Cloud, Python, SQL and Distributed Programming
Experience in NLP, Transformer, LIME, SHAP, ML, BERT, Keras, Spark, FAISS, Elastic Search, Scikit, numpy, pandas, Python, Statistic modeling and DL.
My motto is sharing my experience, best-practices and learning with everyone. Please subscribe and support the


Lang Chain Agent for beginner
21:04
2 месяца назад
Комментарии
@cdm6541
@cdm6541 11 дней назад
Can you please specify in which sequence we have to run these codes to get flask application to work?
@cdm6541
@cdm6541 11 дней назад
I am facing a internal server error on website and in vscode it says unable to store string as float( preprocess review data )
@dev3446
@dev3446 11 дней назад
@cdm6541 please share the code and error details
@dev3446
@dev3446 11 дней назад
To run flask application you just need to run SentimentAnalyzerEngine.py file
@cdm6541
@cdm6541 11 дней назад
@@dev3446 I get this error when I enter sentiment in the flask application website: ValueError: could not convert string to float: " Must watch movie.. Breathtaking, Benchmark in Kannada industry, Story, Screenplay, Casting full marks. Actors Screen presence is mind blowing. Cinematography is remarkable. Hats off to director's imagination. International standard movie" 127.0.0.1 - - [20/Oct/2024 13:30:25] "POST /predict HTTP/1.1" 500 -
@cdm6541
@cdm6541 11 дней назад
@@dev3446 I get this error when I enter sentiment in the flask application website: ValueError: could not convert string to float: " Must watch movie.. Breathtaking, Benchmark in Kannada industry, Story, Screenplay, Casting full marks. Actors Screen presence is mind blowing. Cinematography is remarkable. Hats off to director's imagination. International standard movie" 127.0.0.1 - - [20/Oct/2024 13:30:25] "POST /predict HTTP/1.1" 500 -
@jackq9197
@jackq9197 19 дней назад
Hi Sir, in phase 3 the scripts was working fine until reaching the model_7, however after executing "Evaluation of FineTuned Logsitic Regression Classifier" , I got kernel died and restarted automatically. It's really confusing as it didn't returning any error messages. When I repeat it again from the beginning and I skip model_7, same issue happened for model_8 and model_9. Kindly your comments and how to fix this.
@dev3446
@dev3446 14 дней назад
It’s hardware issue, your system has performance issues. Please reduce range of hyper parameter. Then it’s work for you
@syaifalkhalim4098
@syaifalkhalim4098 20 дней назад
do you not using naive bayes sir?
@dev3446
@dev3446 20 дней назад
This playlist goal is to give you idea how you can develop a ML solution end to end. From Data preprocessing to cloud deployment. Further you can try different models, also generative ai modals as well
@parthivganguly6300
@parthivganguly6300 20 дней назад
Why did you make a new logistic regression model, and didn't used the same one from previous video?
@dev3446
@dev3446 20 дней назад
Please follow the instructions as per videos. Because for Productionize a model we need end to end pipeline. So that, all preprocessing, features engineering steps would be included in pipeline
@cdm6541
@cdm6541 20 дней назад
Brother can you specify how exactly did you deal with the sarcasm, multipolarity and the negation in your project?
@dev3446
@dev3446 20 дней назад
Yes, please refer to EDA video in EDA part you will find hacks to find sarcasm, negation issues
@parthivganguly6300
@parthivganguly6300 21 день назад
It took around 46 mins for ada boost classifier, but for other models my laptop took same time as yours. So why for ada boost model it's taking so much time?
@dev3446
@dev3446 20 дней назад
Ada boost required more computation. That why it take time. First option, You can reduce hyper parameter range to cut training time. Second option high config machine
@badrinath218
@badrinath218 Месяц назад
If anybody is having trouble with loading the dataset use (on_bad_lines='skip' , nrows=150001) in pd.read_csv
@dev3446
@dev3446 Месяц назад
Please share error details
@sivadevil4845
@sivadevil4845 Месяц назад
u made good videos, u deserve more subscribers.
@dev3446
@dev3446 Месяц назад
Thanks for your support
@cdm6541
@cdm6541 Месяц назад
Sir i am facing an issue in vectorization code snippet There is a lookup error in line : X_train_count = countvect.fit_transform(train['Reviews_clean']).toarray()
@cdm6541
@cdm6541 Месяц назад
this is the error message: LookupError Traceback (most recent call last) Cell In[43], line 4 2 countvect = CountVectorizer(analyzer = "word", tokenizer = LemmaTokenizer(), ngram_range=(1,1),min_df=10,max_features=500, lowercase=False) 3 tfidfvect = TfidfVectorizer(analyzer = "word", tokenizer = LemmaTokenizer(), ngram_range=(1,1),min_df=10,max_features=500, lowercase=False) ----> 4 x_train_count = countvect.fit_transform(train['Reviews_clean']).toarray() 5 x_test_count = countvect.transform(test['Reviews_clean']).toarray() 6 x_train_tfidf = tfidfvect.fit_transform(train['Reviews_clean']).toarray() File c:\Users isha\AppData\Local\Programs\Python\Python311\Lib\site-packages\sklearn\base.py:1473, in _fit_context.<locals>.decorator.<locals>.wrapper(estimator, *args, **kwargs) 1466 estimator._validate_params() 1468 with config_context( 1469 skip_parameter_validation=( 1470 prefer_skip_nested_validation or global_skip_validation 1471 ) 1472 ): -> 1473 return fit_method(estimator, *args, **kwargs) File c:\Users isha\AppData\Local\Programs\Python\Python311\Lib\site-packages\sklearn\feature_extraction\text.py:1372, in CountVectorizer.fit_transform(self, raw_documents, y) 1364 warnings.warn( 1365 "Upper case characters found in" 1366 " vocabulary while 'lowercase'" 1367 " is True. These entries will not" 1368 " be matched with any documents" 1369 )
@cdm6541
@cdm6541 Месяц назад
is there an issue with data cleaning or tokenization. i even tried turning lowercase=False but its still there
@cdm6541
@cdm6541 Месяц назад
issue got resolved, i needed to download "nltk.punkt_tab"
@dev3446
@dev3446 Месяц назад
@cdm6541 please share error details
@dev3446
@dev3446 Месяц назад
Please share error details
@saikrishnasirikonda1008
@saikrishnasirikonda1008 Месяц назад
Good to start on LLM Agents! can you provide an info to fetch the reference docs(actual docs where we uploaded) in LLM + RAG
@dev3446
@dev3446 Месяц назад
Thanks 🙏🏻, In RAG system we can upload multiple type of doc through documents loader like pdf, text, markdown etc. for more info please refer to this video Create a Chatbot with RAG and Opensource LLM with Lang Chain and Flask for Beginners ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-LiZLJAUz_7A.html
@arnabrahmanzahid8296
@arnabrahmanzahid8296 2 месяца назад
what is the reason it showing key error 118,whats the solution
@arnabrahmanzahid8296
@arnabrahmanzahid8296 2 месяца назад
Cant it be run directly downloading from github? without data preprocessing steps
@dev3446
@dev3446 2 месяца назад
This project has seven stages from data cleaning to cloud deployment. So, you need to follow all stages
@dev3446
@dev3446 2 месяца назад
Please share more details regarding error and in which module you are facing this error
@saqibramzan3749
@saqibramzan3749 3 месяца назад
Sir one last thing can you tell is part 3 is enough for our mini project as we have to submit project like smester project till part 3 is enough or not mean they meet our all requirements for project sentimenal analysis or not kindly explain
@dev3446
@dev3446 3 месяца назад
We have covered Data analysis and feature engineering in first 3 videos. If that’s fine for your project then you can use only up to 3 parts
@saqibramzan3749
@saqibramzan3749 3 месяца назад
Hi Sir Sir I want to know how we adjust the other language mean other than english how they effect our results and what the purpose of other language please give satisfying answer and long ans so i can better uderstan please ans....
@dev3446
@dev3446 3 месяца назад
If you have any requirements to work on other languages then you can apply same techniques like feature engineering and models. You just need to understand those languages so that you can do better analysis
@saqibramzan3749
@saqibramzan3749 3 месяца назад
@@dev3446 Thanks so much sir
@saqibramzan3749
@saqibramzan3749 3 месяца назад
Hi Sir Sir I want to know how we adjust the other language mean other than english how they effect our results and what the purpose of other language please give satisfying answer and long ans so i can better uderstan please ans....
@dev3446
@dev3446 3 месяца назад
1. If we have project requirements to train ML model then we need training data for particular language. Only training/test data change rest technique would be same as we do data analysis, feature engineering for English language. Please try to understand fundamentals of ML/DL.
@Cosmogyral04
@Cosmogyral04 3 месяца назад
00:02 Deployment of the base model onto the local server 01:34 Introduction to sentiment analysis framework and its modules 02:57 Using pipeline for data preprocessing 04:28 Using SQL pipeline and TFID vectorizer for data cleaning and model deployment 05:53 Deploying machine learning model using Flask for sentiment analysis project 07:26 User input for sentiment analysis prediction 08:57 Deploying sentiment analysis model with Flask 10:47 API feeds data for sentiment analysis and creates a pipeline for base models Crafted by Merlin AI.
@kdpr007
@kdpr007 3 месяца назад
one quick question" @16.54 min why are step 3 and 4 having the same indentation level? In a normal function, step4 is at the same indentation as step 8 : func() Kindly help me understand
@kdpr007
@kdpr007 3 месяца назад
Thank you for the video, a very detailed one. One quick question, why is the return keyword not indented same as print("Below is the nested function")? or in otherwords why is return is the same indentation as def function_returned? Kindly help me understand
@dev3446
@dev3446 3 месяца назад
Thanks 🙏🏻
@kdpr007
@kdpr007 3 месяца назад
@@dev3446 any possible responses to my question?
@dev3446
@dev3446 3 месяца назад
@kdpr007 in step 3 we are defining inner functions and step 4 we are returning values. So, Both are inside the parent function with same level indentation
@Cosmogyral04
@Cosmogyral04 3 месяца назад
00:02 Model evaluation with Shap and LIME 02:27 Model evaluation and corrective actions for bias and errors 04:44 Model evaluation using multiple matrices resulted in high F1 score and accuracy. 07:02 Understanding confusion matrix and feature contributions in prediction analysis 09:19 Analyzing feature contributions for sentiment analysis using Shap values 11:44 Evaluating feature contributions and applying thresholding techniques for sentiment analysis. 00:00 Analyzing feature contributions using Shap and LIME 18:16 Explaining feature impact on predictions using regression models 20:43 Improving model accuracy in real time 23:08 Explaining model evaluation using Shap and LIME for sentiment analysis project Crafted by Merlin AI.
@Cosmogyral04
@Cosmogyral04 3 месяца назад
00:03 Model selection for sentiment analysis. 02:15 Logistic regression showed high precision and low latency 04:23 Decision tree takes significantly longer training time compared to logistic regression. 06:31 Addressing overfitting and accuracy challenges through hyperparameter tuning 08:49 Random forest to AdaBoost comparison and hyperparameter tuning 11:03 Hyperparameter tuning in logistic regression for sentiment analysis 13:13 Hyperparameter tuning for Random Forest Classifier 15:32 Achieving best F1 score through hyperparameter tuning Crafted by Merlin AI.00:03 Model selection for sentiment analysis. 02:15 Logistic regression showed high precision and low latency 04:23 Decision tree takes significantly longer training time compared to logistic regression. 06:31 Addressing overfitting and accuracy challenges through hyperparameter tuning 08:49 Random forest to AdaBoost comparison and hyperparameter tuning 11:03 Hyperparameter tuning in logistic regression for sentiment analysis 13:13 Hyperparameter tuning for Random Forest Classifier 15:32 Achieving best F1 score through hyperparameter tuning Crafted by Merlin AI.
@dev3446
@dev3446 3 месяца назад
Thanks for sharing it
@AnushkaMohan-n8o
@AnushkaMohan-n8o 4 месяца назад
sir lemmatization is taking a lot of time
@dev3446
@dev3446 4 месяца назад
Please let me know your computer configuration details. Seems as hardware issue. As it would be great if you can use i7 gen processor with 16 GB RAM
@AnushkaMohan-n8o
@AnushkaMohan-n8o 4 месяца назад
sir i am not able to download wordnet. it is taking too long. and the code is not working without downloading wordnet.
@dev3446
@dev3446 4 месяца назад
@user-nu1xf9dh6p ok please check your Internet speed and if speed are good and you still facing issues then please share any error details
@dev3446
@dev3446 4 месяца назад
@user-nu1xf9dh6p please check your Internet connection. Because it download gb of data
@anushkamohan887
@anushkamohan887 4 месяца назад
@@dev3446 sir is there any other way i can do lemmatization because i dont think i will be able to download wordnet. The code cell keeps running for a very long time and then it gives an error. i will not be able to send the error to u bcoz it takes very long to try downloading again. although if u want to see it i will try once again. but is there any other way ? I dont know why youtube is not letting me comment so if i can connect to u on gmail or anything else it would be great!
@LOKESHSRINIVAS-x1z
@LOKESHSRINIVAS-x1z 4 месяца назад
Hi, I have tried some chatbot with nemo guardrails but it is not working. could you try and rectify please...
@dev3446
@dev3446 4 месяца назад
Please let me know some detail what kind of chatbot you are build (tech stack) and what kind of issue your are facing
@poetrybets8465
@poetrybets8465 4 месяца назад
Thanks Sir.for sharing RAG workflow details in easy language
@dev3446
@dev3446 4 месяца назад
Thanks for your support
@RanjanKumarMohanty-o6h
@RanjanKumarMohanty-o6h 4 месяца назад
great video❤
@dev3446
@dev3446 4 месяца назад
Thankyou for your support
@avinashkumarbarnwal5912
@avinashkumarbarnwal5912 6 месяцев назад
More informative video
@dev3446
@dev3446 6 месяцев назад
Thanks 🙏🏻
@avinashkumarbarnwal5912
@avinashkumarbarnwal5912 6 месяцев назад
Thank u sir, i completed this project and learn so many thing like chi square test, shap and lime
@dev3446
@dev3446 14 дней назад
Thankyou for sharing your feedback’s
@avinashkumarbarnwal5912
@avinashkumarbarnwal5912 6 месяцев назад
Thank u sir, i completed this project and learn so many thing like chi square test, shap and lime
@dev3446
@dev3446 6 месяцев назад
Thanks 🙏🏻
@shubhamchaudhary1750
@shubhamchaudhary1750 6 месяцев назад
Awesome video explanation ! 🙌
@dev3446
@dev3446 6 месяцев назад
Thanks 🙏🏻
@BiranchiNarayanNayak
@BiranchiNarayanNayak 7 месяцев назад
Nice tutorial
@dev3446
@dev3446 7 месяцев назад
Thanks 🙏🏻
@SANJAYKUMAR-ns4ch
@SANJAYKUMAR-ns4ch 8 месяцев назад
sirr i followed from vid 1 and started vid 2 but the doubt is in which app should i type this code 🥲. I am a complete beginner sir, i don't know like whether should i just type the code directly in vs code and run it
@dev3446
@dev3446 8 месяцев назад
Initially you can start coding for Data Cleaning, Data Analysis, FeatureEngineering and model evaluation in Jupiter notebook. Then you can switch to visual studio code for Model Deployment coding
@SANJAYKUMAR-ns4ch
@SANJAYKUMAR-ns4ch 8 месяцев назад
sir I don't know how to create this project like where should I start coding ?
@dev3446
@dev3446 8 месяцев назад
Please start from video 1 in this playlist. This playlist contains end to end project from data processing to cloud deployment. Just follow the video instructions
@SANJAYKUMAR-ns4ch
@SANJAYKUMAR-ns4ch 8 месяцев назад
Sirr i followed from vid 1 and started vid 2 but the doubt is in which app should i type this code 🥲. I am a complete beginner sir, i don't know like whether should i just type the code directly in vs code and run it 😑@@dev3446
@hellohello-op2np
@hellohello-op2np 9 месяцев назад
sir where in which video you have used NLP
@dev3446
@dev3446 9 месяцев назад
NLP is used in data analysis and feature engineering videos . As we have textual data, so we used NLP library nltk for data cleaning and vectorizer to covert textual data into into vector form
@tavneetmanchanda
@tavneetmanchanda 9 месяцев назад
can you pease tell how to deploy the model from ubuntu to aws
@dev3446
@dev3446 9 месяцев назад
I have shared details steps to deploy a model to Ubuntu instance on AWS in this video ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-921aFe381ro.htmlsi=kj520aeunwNsBgZj
@dev3446
@dev3446 9 месяцев назад
Please refer to this video and let me know if you have any other issue
@tavneetmanchanda
@tavneetmanchanda 9 месяцев назад
HI sir , can you help me with the final part of my project , I want to draw a trend analysis using the sentiment analysis tool , can you please help me , I will be very thankful to you.
@dev3446
@dev3446 9 месяцев назад
Sure, please share more details or contact on +91-9779009307
@FalakSoomro
@FalakSoomro 10 месяцев назад
Amazing keep it up ✨
@dev3446
@dev3446 9 месяцев назад
Thank you 🙌
@tavneetmanchanda5051
@tavneetmanchanda5051 10 месяцев назад
I am getting keyerror 118 , can anyone please help me.. I really need some help.
@dev3446
@dev3446 10 месяцев назад
Please provide more details regarding your error and also share error data
@tavneetmanchanda
@tavneetmanchanda 10 месяцев назад
@@dev3446 After reaching the ui side , when i hit the get sentimnet button I m getting this (NameError: name 'classifier' is not defined) also I am getting the exception /home/tavneet.manchanda/Desktop/Sentiment_Analysis_Case_Study/sentimentanalysis/models/model Error loading pickle file: invalid load key, 'v'.
@dev3446
@dev3446 7 месяцев назад
I think you have pickle model path issue. Please check your directory where you store trained pickle model
@AbcdAhyst
@AbcdAhyst 10 месяцев назад
bhai dataset link can u share please
@dev3446
@dev3446 10 месяцев назад
Please check video description for Kaggle data link
@dev3446
@dev3446 10 месяцев назад
Data link www.kaggle.com/datasets/crisbam/imdb-dataset-of-65k-movie-reviews-and-translation
@nikhilsastry6631
@nikhilsastry6631 Год назад
Can you make a video on credit fraud detection project & flask app building & deploying it to pythonanywhere website ... Please it would help me as i am beginner in ML
@dev3446
@dev3446 Год назад
Sure I will try to create videos related to Fraud Detection. Further, Incase of Credit Fraud detection only Feature Engineering and Modelling part would be changed, Rest Flask App building and development steps would be same.
@jiaali6565
@jiaali6565 Год назад
@dev3446 Is the deployment part on AWS free?
@dev3446
@dev3446 Год назад
Yes, we deployed small model using free tier EC2 instance
@YashSharma-le3mo
@YashSharma-le3mo Год назад
Hi sir Can you give idea for. My final year project on sentiment analysis on which research paper can be published
@dev3446
@dev3446 Год назад
You can try different ML models and ensemble modes for Sentiment Analysis and also you can use LLM for Sentiment Analysis project.
@harsh2014
@harsh2014 Год назад
Nice Discussion !
@ramshrivastava6333
@ramshrivastava6333 Год назад
Where is kaggle link??
@dev3446
@dev3446 Год назад
Here are the link: www.kaggle.com/datasets/crisbam/imdb-dataset-of-65k-movie-reviews-and-translation
@orthodoxreview8490
@orthodoxreview8490 Год назад
Sir, why didn't we do lemmatization?
@dev3446
@dev3446 Год назад
We have done lemmatization in processing pipeline. Please refer to class LemmaTokenizer class under dataprocessing.py module.
@ankitkr8223
@ankitkr8223 Год назад
Sir what if I input plain text for checking sentiment.. does the model still give the outcome as positive and negative or it's just for movie reviews?
@dev3446
@dev3446 Год назад
Yes, Ankit it give you predictions for any kind of text. Because in the pipeline text convert into embedding. Embedding is used by model for prediction. But as model trained on movie reviews data. So, it would provide better predictions if text contains similar features(words)
@_YashBangera
@_YashBangera Год назад
sir can you use this code to get an output like : you enter the movie name and the system tell you if the movie is good or bad ?
@dev3446
@dev3446 Год назад
Sure you just need to call movie name and label
@dev3446
@dev3446 Год назад
You do it easily
@_YashBangera
@_YashBangera Год назад
15:51 can I skip the assignment and still get the same results?
@dev3446
@dev3446 Год назад
Yes no issue you can skip assignments
@_YashBangera
@_YashBangera Год назад
15:59 why so much hours bhai?
@dev3446
@dev3446 Год назад
Hi Yash please share more detail what’s your question
@dev3446
@dev3446 Год назад
Incase of any issue you can call me or what’s me on this number +91 97790 09307
@aeroxec7918
@aeroxec7918 11 дней назад
​@@dev3446sir i send a *msg* to you on a issue please check
@sudeeppundir8206
@sudeeppundir8206 Год назад
Nice
@dev3446
@dev3446 Год назад
Thanks for your support
@sudeeppundir8206
@sudeeppundir8206 Год назад
Nice
@dev3446
@dev3446 Год назад
Thanks for your support
@hitesh1779
@hitesh1779 Год назад
can a beginner can do these project
@dev3446
@dev3446 Год назад
Yes Hitesh if you know python and basic of ML. Then you can easily do this project. Further I have provided all code and GitHub link in description. Incase of any issues you can directly contact Me
@hitesh1779
@hitesh1779 Год назад
@@dev3446 Ya I know the basic pyhton, ill try these project Thank u for your guidance
@dev3446
@dev3446 Год назад
Sure
@saqibramzan3749
@saqibramzan3749 3 месяца назад
@@dev3446 how we can directly contact you sir?
@anabiya3587
@anabiya3587 Год назад
Thank you so much, it's really very helpful video! 😍Could you please provide the GitHub link and data set link?
@dev3446
@dev3446 Год назад
Thanks Ana for your support. 1. Github for repository for Sentiment Analysis Project: github.com/OptimalXAI/Sentiment_Analysis_Case_Study.git 2. In case of Python Playlist: you can find google Colab link in every video description Further, we will add readme in Github Repository to install all prerequisite dependencies easily by tomorrow.
@anabiya3587
@anabiya3587 Год назад
@@dev3446 sir, You're welcome. Okay sir, thanks a lot for your reply.
@dev3446
@dev3446 Год назад
I have added readme. Now you can easily use this project
@anabiya3587
@anabiya3587 Год назад
@@dev3446 okay sir, thanks a lot for your kind response. Best wishes for you and your channel. 🏆
@dev3446
@dev3446 Год назад
Thanks for best wishes