After watching most of your videos, I only can say that you're a great teacher, a good guide for the people who are struggling to understand concepts of Data science. Keeping sharing and blessing us.
Hello My name is Cleber and I'm from Brazil. Thank you very much for this excellent lesson (step by step). I couldn't find any video that explained me in detail, including making the prediction showing the probability. Stay in Peace.
Thank you Thank you Thank you so much Sir. I really don't have enough words to thank you. You have made this video because I had requested you. Why I didn't ever met a Professor like you who is extremely helping humble and kind towards their students. With due respect I'm thanking you for making this video.
Thank you so much for the videos! I always follow your videos and learn something new. I have a quick question about the math. You use P(A) amd P(B) for demonstration, can you also write down the math for all 4 variables? Bayesian rule for P(A), P(B), P(C), P(D) in one or multiple formula? Thank you so much. Allen
Sir, at 13:20, you are using the model and training data set to predict.when we have used the train data to model the data set, what is the reason to do prediction using the same train data on the model ?
We do it for both train and test data to make sure the results are consistent. If train data results in accuracy of 90% but the test data has only 20%, then we will know that we are doing over fitting.
Thanks for the video- just started learning with R so this is definitely useful for me! Qns: how did u train the data? Was it just by keying in those codes before u applies Naive bayes?
Great work and detailed tutorial. It worked perfectly on different data i used. Please sir how to one predict a single data after using the model. The model was been trained and working well. How can we predict a single data into the model and see the outputs. I would love to have your email sir.
Many thanks sir for this video. A few quick questions though. #In the case where there exists a high correlation between independent variables (ie multicollinearity), how do I treat it to make it suitable for the application of Naive Bayes in the sense of classification ? #What characteristics must one look up for in a data to know the suitability of applying Naive Bayes ? # May you please explain what this code means "data %>% ......" in plotting the box and density charts ? Thanks once again for this video.
PCA can help with multicollinearity or you can go with ridge, lasso or elastic net regression. When developing classification model, generally you should try different methods and see which one works best with the data you have. Each data has it's own specific characteristics and it is not possible to know beforehand which method will perform better. For %>% explanation, see this link: th-cam.com/video/niB5A8qa88I/w-d-xo.html
Thanks in advance sir Could you please clear my query that in P1 & P2 I am getting same length of arguments. I have given test data to predict P2 but still getting same output as P1 & their by getting error at tab2 as " all arguments must have the same length ". Please rectify Thanks again.
You should always pass the model as the first argument in predict function. The second parameter should be a data frame of predictor variables only. You can specify type=”prob” as an extra argument to get probabilities of every factor of y. Either type=”class” directly gives you the class of predicted values. By default type argument is set up differently for every R version.
Dr., one more question. I wish that in the classification on the test basis, I could set the threshold percentage to perform the classification. Example: There is a 70% chance of being in class 1 and 30% in class 2. I would like it to be classified as class 2 odds greater than 10%. How to set this threshold for classification. Is it in the predict function? Thanks!
Hello sir, question about this model. if you perform the naive_bayes function to predict a column and do not use all independent variable columns in the test data should you also exclude these columns when performing the prediction?
Thank you so much for the videos! Dr. Rai, I tried this at different dataset. But I have a problem result on modelling. there is "# ... and 8 more tables". How can I show full of result?
They are factor variables. We should not have combinations of the two where frequency is zero. If it is zero or too small, we can probably combine rank-3 and rank-4 into one.
you have been an inspiration for me to get into R world.. pls, let me know how to start to R with the basics to make a career in it. Hoping for ur positive reply.
You can start with following playlist: th-cam.com/play/PL34t5iLfZddtII4ssT8FSUFP27fPYDEhY.html In addition, it will help to review statistical concepts: th-cam.com/play/PL34t5iLfZdduw0BfhRbNXxi_Y3BGk2MoB.html
Sir, I would like to know whether we can consider the Submission Status as Dependent attribute for this above method. Whereas the attribute contains "Available profile", "Level 1", "Level 2", "Placed". Hiring process by the Recruitment team. Need to find which team is hiring best profile and classifying them into group as "Tier 1", ""Tier 2" and "Tier 3". Plz, suggest your opinion on this...
Ive tried different data set but on the plot (Model) im getting an error of " Error in xy.coords(x, y, xlabel, ylabel, log) : 'x' is a list, but does not have components 'x' and 'y' " what should i do?
I want to predict KIDNEY disease using naive bayes classifier in Rstudio.So I want to know will your R script works for it or not? If not what should I do?
It depends on the data and type of variables you are using. Generally you should not try just one method, but several and then see which one does best.
to install psych package, you can run install.packages('psych') You can follow similar process for other packages. After installing a package, you should run library line.
Sir, what am I doing wrong here? model= naive_bayes(admit~., data= train) p= predict(model, train, type = 'prob') Error in which((sapply(newdata[ind_factor], nlevels) != sapply(tables[ind_factor], : (list) object cannot be coerced to type 'integer'
hello sir, the video is very useful, thank you so much, can I know the version of R software we use to download bcoz presently I have version4.0 and in that pairs.panel package is not available so kindly inform me about the version of R software that has to be downloaded
I just downloaded 4.0 and then installed 'psych' again. Initially pairs.panels didn't run as you mentioned. However, after I ran library(psych) and retyped pairs.panels command, it worked fine.
Tq for this video it was good and easily understandable than others... Naive bayes classification method is applicable for fraud detection in creditcard transaction dataset because it contains 30 k rows of data with 30+ columns sir. Please reply me sir i have this confusion from long time...
You have a wonderful talent to teach. You are one of the best that I have seen. When I execute the program I get the following error in the predict, “> p p head(cbind(p, train)) Error in cbind(p, train) : object 'p' not found” Can you provide a solution?
Hello sir, hope you are doing good! i am writing you because I came across a paper named Yagyanath Rimal ,with your naive bayes analysis as same as in this video. The journal name is INTERNATIONAL JOURNAL ON ORANGE TECHNOLOGIES (IJOT) with paper title Naïve Bayes Machine Learning Classification with R Programming: A case study of binary data sets, please go through if they have used your analysis with their name
@@bkrai rank would seem to be categorical, and admit binary/dummy. I guess I'm missing something about how R defines factor variables. Does it just include these types of variables? Thanks!
When a variable such as admit has values like 0 and 1, it thinks it is a integer variable. Similarly rank has values such as 1, 2, 3, and 4. This is also treated as integer by default. In such situations we need to make changes where necessary.
Error in data[partition_data == 1, ] : object of type 'closure' is not subsettable getting this error here is the link to the code: - github.com/JordanTheDodger/NaiveBayes-in-R-Test- I looked on StackOverflow, it said I am trying to use some inbuilt function. Thanks in Advance. Looking forward to your reply
![บุษบา - เมนทอล [Official MV]](http://i.ytimg.com/vi/IGnWPokSEis/mqdefault.jpg)
![มองหน้ากันไม่ติด (Awkward) - โอ๊ต ปราโมทย์ FEAT. MAIYARAP [Official MV]](http://i.ytimg.com/vi/4b02SY_r6Gw/mqdefault.jpg)
After watching most of your videos, I only can say that you're a great teacher, a good guide for the people who are struggling to understand concepts of Data science.
Keeping sharing and blessing us.
Thanks for your comments!
I have this to say. You are such a great teacher. Very apt and intuitive video. I must confess your videos are one of the best.
Thanks for your comments!
Hello
My name is Cleber and I'm from Brazil. Thank you very much for this excellent lesson (step by step). I couldn't find any video that explained me in detail, including making the prediction showing the probability. Stay in Peace.
Thanks for comments!
Your Naive Bayes video is soooo much better than what I receive during my graduate lectures.
Thanks for your feedback and comments!
Your videos are the best! Thank you for your efforts and passion for educating your fellow beings!
Thanks for your comments!
Your videos couldn't be better & simpler! You help us alleviate from the fear of data science & machine learning! Thanks a ton sir! And bless us!!!
Glad to hear that!
hi do you have the data file
The demonstration & explanation of the viz. part is highly appreciable.
Thanks for comments!
Thank you so much, Dr Rai. For your clear and detailed explanation, that is really helpful!
You're most welcome!
Thank you Thank you Thank you so much Sir. I really don't have enough words to thank you. You have made this video because I had requested you. Why I didn't ever met a Professor like you who is extremely helping humble and kind towards their students. With due respect I'm thanking you for making this video.
I have no words Really its amazing sir
Thanks for comments!
Thank you for making this video it is extremely helpful!!
You're so welcome!
Thank you very much for your helpful and insightful videos
You are very welcome!
Thank u sir useful video
Thanks alot
Most helpful for naive bayes
Learned alot
Thanks for comments!
I'm enjoying your video every'time!!!
Thanks for comments!
Thanks for the video.. Very useful
Thanks for comments!
روووووعة يسلمو .... Thaaaaanx alot
You are welcome!
It helped me a lot. Thank you. Very powerful explanation.
Thanks for your comments!
@@bkrai Absolutely. We are trying to analyze your video now.
Thanks!
thank you.. sir, can we use naive bayes classification for sentimental youtube comments... if possible can you please upload a video on that..
Liked and subscribed because my guy, Bharatendra, is the truth! Thank you!
Thanks for comments!
Thank you so much, I really need that
Thanks for comments!
I like it , thank you for sharing ….. enjoy your life
Thanks for comments!
This was great, thank you.
You're very welcome!
Thank you so much for the videos! I always follow your videos and learn something new. I have a quick question about the math. You use P(A) amd P(B) for demonstration, can you also write down the math for all 4 variables? Bayesian rule for P(A), P(B), P(C), P(D) in one or multiple formula? Thank you so much. Allen
Thank u
I like your vidss. Thank you.
Thanks for comments!
I don't know who the hell unlike ur videos.... ur videos are always awesome
Thanks for your comments!
which version of R studio you r using ?
Sir, at 13:20, you are using the model and training data set to predict.when we have used the train data to model the data set, what is the reason to do prediction using the same train data on the model ?
We do it for both train and test data to make sure the results are consistent. If train data results in accuracy of 90% but the test data has only 20%, then we will know that we are doing over fitting.
Hi when i used your model to plot the density plot my graph was completely different , any solution please
Thanks for the video- just started learning with R so this is definitely useful for me! Qns: how did u train the data? Was it just by keying in those codes before u applies Naive bayes?
In line 32 when we develop a model, it is done after training.
Sir can we use naive bayes classification for sentiment analysis of tweets, could you please do a video based on that??
Thanks for the suggestion, I've added this to my list.
you save my life bro
Thanks for comments!
Great work and detailed tutorial. It worked perfectly on different data i used. Please sir how to one predict a single data after using the model. The model was been trained and working well. How can we predict a single data into the model and see the outputs. I would love to have your email sir.
thanks!!! friend.
thanks for comments!
Many thanks sir for this video. A few quick questions though.
#In the case where there exists a high correlation between independent variables (ie multicollinearity), how do I treat it to make it suitable for the application of Naive Bayes in the sense of classification ?
#What characteristics must one look up for in a data to know the suitability of applying Naive Bayes ?
# May you please explain what this code means "data %>% ......" in plotting the box and density charts ?
Thanks once again for this video.
PCA can help with multicollinearity or you can go with ridge, lasso or elastic net regression. When developing classification model, generally you should try different methods and see which one works best with the data you have. Each data has it's own specific characteristics and it is not possible to know beforehand which method will perform better. For %>% explanation, see this link:
th-cam.com/video/niB5A8qa88I/w-d-xo.html
Many thanks DR RAI
Thanks in advance sir
Could you please clear my query that in P1 & P2 I am getting same length of arguments.
I have given test data to predict P2 but still getting same output as P1 & their by getting error at tab2 as
" all arguments must have the same length ".
Please rectify
Thanks again.
You should always pass the model as the first argument in predict function. The second parameter should be a data frame of predictor variables only. You can specify type=”prob” as an extra argument to get probabilities of every factor of y. Either type=”class” directly gives you the class of predicted values. By default type argument is set up differently for every R version.
Thank you very much. I believe there is an error at 6'43''. A and B are not independent. What is independent is the features B1, B2, ...Bn.
Dr., one more question. I wish that in the classification on the test basis, I could set the threshold percentage to perform the classification. Example: There is a 70% chance of being in class 1 and 30% in class 2. I would like it to be classified as class 2 odds greater than 10%. How to set this threshold for classification. Is it in the predict function? Thanks!
You can try this link: th-cam.com/video/ypO1DPEKYFo/w-d-xo.html
Hello sir, question about this model. if you perform the naive_bayes function to predict a column and do not use all independent variable columns in the test data should you also exclude these columns when performing the prediction?
Use only those variables while developing the model that are also part of test data.
where can i get the binary.csv file to practice ,let me know
See the link in the description area below this video.
Sir can you explain how to create a multiple output neural network in r using neuralnet package?
You can use following link for multiple outputs. This is much faster.
th-cam.com/video/hd81EH1g1bE/w-d-xo.html
GOOD DAY, SIR! i would like to ask how to train and test the data after sentiment analysis
Thank you so much for the videos!
Dr. Rai, I tried this at different dataset. But I have a problem result on modelling. there is "# ... and 8 more tables". How can I show full of result?
Let me know code that you are referring to.
It was very helpful..sir,I have a question can we do clustering using naive Bayes..If yes, then how can we do it??
For clustering I would suggest these 6 videos:
th-cam.com/play/PL34t5iLfZddvMPAl1TzHJ_GjQcD3s6w_Z.html
thanks sir, query= cross-validation why we need to check between admit and rank, you talked about frequency and i didn't understand ?please clarify..
They are factor variables. We should not have combinations of the two where frequency is zero. If it is zero or too small, we can probably combine rank-3 and rank-4 into one.
sir, firs hats-off to you for the swift response and now i understand the concept
you have been an inspiration for me to get into R world.. pls, let me know how to start to R with the basics to make a career in it. Hoping for ur positive reply.
You can start with following playlist:
th-cam.com/play/PL34t5iLfZddtII4ssT8FSUFP27fPYDEhY.html
In addition, it will help to review statistical concepts:
th-cam.com/play/PL34t5iLfZdduw0BfhRbNXxi_Y3BGk2MoB.html
@@bkrai Thanku so much sir
Welcome!
Sir, I am getting an error , "variable lengths differ" can you please assist!!
In case it is due to missing values, then you can use thins link: th-cam.com/video/An7nPLJ0fsg/w-d-xo.html
Helo Sir, Can I kindly know the version of R software used in this vedio.
This video was done around Feb 2018. So whatever version was available at that time.
Sir,
I would like to know whether we can consider the Submission Status as Dependent attribute for this above method. Whereas the attribute contains "Available profile", "Level 1", "Level 2", "Placed".
Hiring process by the Recruitment team.
Need to find which team is hiring best profile and classifying them into group as "Tier 1", ""Tier 2" and "Tier 3".
Plz, suggest your opinion on this...
In the box plot line (the whole pipelining) , I am getting an error stating ( Could not find function "+
You mentioned Could not find function "+
Hi, Can we apply Naive Bayes on a data that has a class imbalance ? This data has a class imbalance . And this is not solved
Yes, it is better to address class imbalance first.
sir, Do u have made any video on EFA, CFA and SEM for research purpose. If yes pls let me know. Thanks in advance.
Not at this time, but I've added them to my list.
great ! i just subscribed boss
Thanks!
That's perfect example. It made really easy to understand! Could you please help us sharing python code for the same example..
Thanks in advance :)
I don't have python code now, but will explore during summer.
@@bkrai Thank you so much! That will be really helpful if you can share as soon as you can.. :)
Ive tried different data set but on the plot (Model) im getting an error of " Error in xy.coords(x, y, xlabel, ylabel, log) :
'x' is a list, but does not have components 'x' and 'y' " what should i do?
Try to convert them to a data frame, because it says 'x' is a list.
great !!!
Thanks!
I want to predict KIDNEY disease using naive bayes classifier in Rstudio.So I want to know will your R script works for it or not?
If not what should I do?
It depends on the data and type of variables you are using. Generally you should not try just one method, but several and then see which one does best.
Sir,is there any general comment regarding,which is better classifier
Naive bayes or logistic regression?
It depends on the data. It is better to try both and choose the one that gives higher accuracy.
how to make the naives bayes model when the target is quantitative?
It is mainly used as a classifier with qualitative target. For quantitative target, try this:
th-cam.com/video/_3xMSbIde2I/w-d-xo.html
How can I install nativebayes library.
First you want to enter the code given below:
install.packages("naivebayes")
library(naivebayes)
hie sir!!! i ran this model and in the console it is not showing all the tables instead it says 2 more tables.How to view those tables as well???
You can maximize the console screen to see everything.
At what time point in the video are you referring to?
sir, I have a question. How can we improve our accuracy, if we come to know our model is bad ? #Thank_You_For_Your_Lecture.
Usually we try more than one model and select one that provides better accuracy.
Sir, Please make a video on Laplace smoothing in R software
Thanks, I've added it to my list.
Teacher, I tried to type follow you but still error. Why? 😢 can you tell me how to install package.
to install psych package, you can run
install.packages('psych')
You can follow similar process for other packages. After installing a package, you should run library line.
When I do classification, furing the test with a test sample everything suddenly becomes "FALSE" anyone have any idea?
Sir, what am I doing wrong here?
model= naive_bayes(admit~., data= train)
p= predict(model, train, type = 'prob')
Error in which((sapply(newdata[ind_factor], nlevels) != sapply(tables[ind_factor], :
(list) object cannot be coerced to type 'integer'
i get the same error dude.. can someone help us rectify the error!!
sir i will provide dengue disease data set u predict the accuracy by naive bayes algorithm please
I saw this today. You can send data at seemabharat@gmail.com
How to take a specific test? Example: I want to know the probability of gre = 545, gpa 3.50 and rank = 3.
On way is to use these lines: z
hello sir. with the code that you use to partition the data, how does it partition? is it by random?
Yes it's random.
@@bkrai Thank you.
You are welcome!
sir i m getting error in installing packages of naviebayes and psych
Check spelling of the package.
where can we get the dataset?
There is a link in the description below this video.
hello sir thank you for all
i have a question why did not use boruta package for signification feature
To avoid any confusion, this video just focuses in naive Bayes.
@@bkrai 👍👍👍👍
so do we have to
No harm in trying. If it helps to improve accuracy, then should use. If accuracy dosen't improve, then can be ignored.
@@bkrai thank you my teacher you are the best
You are very welcome!
hello sir, the video is very useful, thank you so much, can I know the version of R software we use to download
bcoz presently I have version4.0 and in that pairs.panel package is not available so kindly inform me about the version of R software that has to be downloaded
I just downloaded 4.0 and then installed 'psych' again. Initially pairs.panels didn't run as you mentioned. However, after I ran library(psych) and retyped pairs.panels command, it worked fine.
@@bkrai yes sir ..I too tried thank u so much sir
sir i have some doubts regarding applying this model to my data set, can I get your email id for suggestions
Thanks for the update!
This is awesome. More samples would also improve the model performance wouldn't it?
Could you please make a video on sentiment analysis?
You are right! Also here is the sentiment analysis video: th-cam.com/video/otoXeVPhT7Q/w-d-xo.html
@@bkrai hello after compiler libraries it's telling there is no such packages .
Please help me I am new to R
Make sure you also run library line after installing packages.
@@bkrai can I know how to install those packages
Tq for this video it was good and easily understandable than others... Naive bayes classification method is applicable for fraud detection in creditcard transaction dataset because it contains 30 k rows of data with 30+ columns sir. Please reply me sir i have this confusion from long time...
Yes, it should work fine.
@@bkrai Tq sir
You have a wonderful talent to teach. You are one of the best that I have seen.
When I execute the program I get the following error in the predict,
“> p p head(cbind(p, train))
Error in cbind(p, train) : object 'p' not found”
Can you provide a solution?
Can you kindly send me the source code to crupasinghe.elite@gmail.com to troubleshoot my problem? Best, Chamila
Hola ,podría compartir la data?
See link in description area.
Do you do any private tutoring?
If you have any query regarding any video here, you can always post it and I'll try my best to provide feedback.
Hello sir, hope you are doing good! i am writing you because I came across a paper named Yagyanath Rimal ,with your naive bayes analysis as same as in this video. The journal name is INTERNATIONAL JOURNAL ON ORANGE TECHNOLOGIES (IJOT) with paper title Naïve Bayes Machine Learning Classification with R Programming: A case study of binary data sets, please go through if they have used your analysis with their name
Thanks, I'll look into it!
Why did you convert from integer to factor?
Rank and admit are factor variables, so they need to be converted to factor.
@@bkrai rank would seem to be categorical, and admit binary/dummy. I guess I'm missing something about how R defines factor variables. Does it just include these types of variables? Thanks!
When a variable such as admit has values like 0 and 1, it thinks it is a integer variable. Similarly rank has values such as 1, 2, 3, and 4. This is also treated as integer by default. In such situations we need to make changes where necessary.
Amazing explanation. I still have one doubt. Shouldn't we pass test data as the required data in predict function?
That's what is done in line 52.
sir can you please explain the cross tabulation between admit and rank. what we want to observe cross tabulation?
Number in each cell should be 5 or more.
you should inform that those libraries had a lot of dependencies.
Error in data[partition_data == 1, ] :
object of type 'closure' is not subsettable
getting this error
here is the link to the code: - github.com/JordanTheDodger/NaiveBayes-in-R-Test-
I looked on StackOverflow, it said I am trying to use some inbuilt function. Thanks in Advance. Looking forward to your reply
Link to the code doesn't work!
Hi Sir, this video is a great! But I can't seem to be able to plot(model), the histogram doesn't seem to show up
make sure there is enough space for a plot to show up in the 4th window.
And then I want dataset for any one Tamil channel like star Vijay, suntv ,zee Tamil this is my humble request sir
Unfortunately I do not have data about Tamil channels.
Bharatendra Rai k sir any sports channel