Dear All, if you are looking for feature engineering materials, please check my feature engineering playlist, all videos are available. Happy Learning!
Sir, Can you please send me the all feature engineering technique file. it will be very helpful to me, if you send them. My email id is ara007kumar@gmail.com
What a coincedence, today is also an Independence day, this really suprised me, I was following your youtube videos and suddenly you greeted, for a movement it got a smile on my face. Happy Independence day.
thank you so much respected sir. Alot love for you from pakistan. this video was very helpfull. we are looking foreword to see others playlist like these from you. once again thanks
thanks sir for listening to my request to create a video on mean encoding , i am really enjoying your videos , and i have learned a lot from that. Please continue to create such awesome videos.
Sir, please share the link once again. I saw your video and it's a very helpful for the student's like me. I want to know more about the feature engineering. Thank you for making such an amazing lecture. Waiting for the feature engineering link.
@@SanthoshKumar-dk8vs you can fork it from either mine or krish's GitHub account. Check Krish's video description for his GitHub link and you find all there
Hi Krish, it is just a suggestion if u start same channel in Hindi language. It will more helpful to those Indian students who are living in small cities and not much familiar with English lecture. Hope u understand my request. I'm your regular viewer and respect ur effort and knowledge. God luck.
Hello sir...your way of teaching is really incredible. I am studying through your lecture for past 1week and that's why unable to fill the form to get the materials which you have prepared for the same... So if possible please enable the form link again...
so happy I found your channel...wooh amazing lecture Please send me the zip file with respect to feature engineering thank you sir will definitely join your channel.
@16:18 position you are saying to use 'one-hot encoding with multi-category' for an ensemble technique. But the beginning of the video you had explained ensemble techniques does not require feature scaling. Can you please clarify?
Thank You Sir!!things we can understand easily by your Videos.Sir could you pleasee reopen the link where we could get the Feature engg materials that could be more great
In mean encoding,If the feature values are replaced by the mean values ,the no of data values in the pincode column are still the same right??Then whats the point of doing mean encoding?
Hi Krish, At 20:23 the Label for A - 0 and A - 1 will be different based on mean right ? for example the mean will be calculated this way right ? A - 1 => 0.73 B - 1 => 0.6 C - 1 => 0.4 A - 0 => 0.5 B - 0 => 0.35 C - 0 => 0.36 Then the ordering of feature will be as below right ? A - 1 >> B - 1 >> A - 0 >> C -1 >> C - 0 >> B - 0
Hi krish, thanks so much for shedding light on this topic of Feature Engineering. I'm at Beginner Level of learning DS/ML and I really fell in love with your way of teaching these techniques. I would really love to get that document on FE you mentioned about in this video. I tried to drop my details via the google form but I see it's closed. Kindly assist please. Thanks in advance!
Sir, I am working in Data Science for a long time but want to your all playlist as I already have covered some of them. I need your notes on Feature Engineering so can you provide me it now. I shall be very thankful to you for this kindness. Best wishes more love for you from Pakistan.
Krish naik you have been so helpful to me on this journey to datascience. The form is no more available is there another way for me to get the feature engineering and again about the Patreon are you still running it or it's already closed?
Hi Krish, I have started liking your channel so much. Hats off for the great service you are doing for the aspiring and already experienced Datascientists. The form url which you have shared is no more available. Could yuo please share the material via google drive or reactivate the form.
by introducing a higher number to the categories on the basis of a higher no. of occurrence in a given class ( say here 1) are you not introducing bias in the dataset? ( target guided ordinal encoding)
Hi Krish ..... I had recently started to follow up your video & it was very helpful. could you please provide me the materials related to feature Engineering......thanks in advance
Hi Krish, I have a question. So, we are normalizing all numerical fields which will make it in the range 0 to 1. While encoding categorical variables, we "one hot encode" a few which will make them either 0 or 1. However, there are a few categorical fields where we are "label encoding" as order matters. It would be encoded like 1,2,3,4,.... which is not in the 0 to 1 range unlike the first and second case that I had discussed. I would like your advice here whether I need to normalize after label encoding the data and range of this case alone won't be in 0 to 1 and normalizing would make it in 0 to 1 range.
Hi Krish, I am confused with your explanation. My doubts are: You said for target guided ordinal categories you are assigning the rank based on mean values then how does it matter if the category is nominal or ordinal since the ranks are assigned based on mean values and not the inherent rank/order of the variable itself. Also for label encoding the number actually mean anything since it isn't like price/sales where the number holds significance so won't the result be junk/unusable and if they are usable then how do you interpret the result
If the categorical feature is ordinal then we can assign labels to it. Here the category which obtains highest target mean will be assigned the highest label value. If the categorical feature is nominal then we cannot assign labels. This is because in label encoding the number does mean something. Different numbers teach the model to make different predictions. For example, a value of 4 (PHD) in the salary prediction example results in a prediction of higher salary whereas a value of 1 (Bcom) results in prediction of a lower salary! This happens because, in the training data, entries with PHD as educational attribute will have a higher salary in the target column. This is generally useful when we do not know the exact ranking!! We find the correlation on the categorical feature to the target and then rank categories according to the mean of the values observed in the target! If the categorical feature is nominal, we do not want the algorithm to learn more from some categories compared to the others. Hence using one hot encoding, we set the values to be 1 and 0. Now as for Mean encoding of nominal categorical features, we essentially map a relationship between the categories and the target. When applied to the example of pin code numbers, some pin codes may result in higher salaries (assuming we are trying to predict salary again). Hence the mean of that pin code will be higher. We simply map in the mean value in place of that category!! So now when the model learns, it will know that a data point with high value in the pin code feature should predict higher salary! Regards!
Hi @Krish Bro Your link for google form is no longer available. And I must thank you for the amazing knowledge you share with us. God bless you for your good work. Trust me, I think you are the only person who can teach data science to anyone, I mean you are making it halwa for your viewers :D. I have watched almost all of your Videos. Thumbs up to you (y) .
Amazing and easy to follow explanations! Newly subscribed and loving it! Just curious how you recover the particular categories and make sense of your results if you use something like mean encoding. Do you have to trace back the original definitions for each mean and what happens if there are repetitions?
Hi sir. I have started to learn ML from your channel only. Thank you for your knowledge that you are sharing with us. I also have one request for you can i get feature engineering zip file now. I am really interested in ML.
Hey Krish! I have few questions based on encoding which are 1. Let’s suppose I have a feature which has 1000 different categories which I need to convert to either an integer/float how should I do that?Here I can’t go for one hot encoding as it might create 999 columns.And also it has only 1100 record /rows by which even though going by the “one hot encoding with multiple categories” method the most repeated categories will be extremely less how do we handle it in such cases? 2. In Ordinal encoding why are the ranks need to be assigned to a categorical label instead we can give some random unique number to the categorical values without ranking them for example PhD as 1 , BE as 2, Masters as 4 and Stats as 3. 3. Also regarding “Label Encoding” how are the ranks decided say if PhD needs to be given higher rank let’s suppose ‘4’ how can a library know that it should be given a higher rank? Or is it something else that in the library code we need to manually set it?
For the first point u should not apply one hot encoding instead we can go ahead with Mean encoding. Label encoding for ordinal categorical will be assigned with ranks. In this case PhD should have a highest rank or label. This will help us to specify the ML algorithm where in we are providing higher importance to phd
Hey Krish, can you please do share the ZIP file which you have mentioned in the video about the Feature engineering, as I am unable to open the Google url link. it will be more helpful if you help me with the file.
Sorry for the stupid question: what´s the output? I mean you encode before you apply any ML algorithm and to that point you just have the dataset, what kind of output do you mean here? Thanx
I agree with all the encoding you treated, however I got doubts on assigning an ordinal label based on the mean value of the output. By doing this you are setting the distance between levels to one, while in reality, based on the output mean, some level might be close than others, with distances less than or more than 1. What do you think? I would also point out the risk of data leak in case of mean encoding, for this reason I'd add a random noise to the mean
In Target guided encoding if mean of two variables are same then how to assign numbers? As both has same mean how to decide for which has to give more numbers.?
This is very informational video. I would like to go through the feature engineering material mentioned here. Will it be possible for this material to be available?
Sir I have a dataset 5000 rows and 36 columns. It is a mixture of both string and interger values. There are are two columns Name and Company which should be encoded. Which method would be the best?
Thank you so much Kris, your videos have been instrumental in my learning process. However, I wanted to find out if it were too late to get the feature engineering code. As it mentioned, the link was opened fir two days hence its been over a year. Or if you have it on your git, I can get it too. I've been struggling with feature engr, though I'm still learning, having that code would mean alot to me. Thank you so much in advance
@@sravanijammula573 Krish uploaded the form when he uploaded the video. Now it's old so I think he removed that. I am also not able to fill the form as I saw video very late
Hi Krish, thanks for this very good video. Just to clarify a bit, when you say Pincode, are you referring to the postal or zipcode? Because from where I am, when they say pincode it means bank's pass code or something like that.
How would we calculate the mean if the output is a multi class classification. In that case shall we take 0,1,2 as output? In the eg you have taken 0&1. Here we can do the calculation. What if there are more than 2 classification outputs. If you could attach the notebooks in a link it would be easier i guess instead of sending personal emails. Just a suggestion.
Sir please could you please tell us why the theory of computation is actually used and what are the application of these subjects please Sir make a video on that
Hi Krish, We cannot perform Mean or Target encoding on test data because we don't have target column in test data. So how can we deal with such a situation where we have variable with multiple level in it? I am talking in respect with Hackathon where we generally don't have target variable, this is something which we have to predict. Would appreciate your help.
You already got the ordinal number or the float number for each category class from the training data . So you dont need to do it again in test data. You will simply use it. You might already know this. But I am answering if someone else has this doubt.
![Noob To Max With DRAGON REWORK In Blox Fruits [FULL MOVIE]](http://i.ytimg.com/vi/LnBMOinoOvA/mqdefault.jpg)
Dear All, if you are looking for feature engineering materials, please check my feature engineering playlist, all videos are available. Happy Learning!
if you don't mind will u reopen the link or provide your writen codes on github with link
@@yash20december all materials are available in feature engineering playlist
Thank you sir.
is there something more you provide for the paid ones. please let me know.
Sir, Can you please send me the all feature engineering technique file. it will be very helpful to me, if you send them. My email id is
ara007kumar@gmail.com
What a coincedence, today is also an Independence day, this really suprised me, I was following your youtube videos and suddenly you greeted, for a movement it got a smile on my face. Happy Independence day.
you are the best, greetings from an ecuadorian studying in Portugal.
There was doubt from so long about this that when there are more than 100 types of value then how to do encoding which is clear today thank you sir 🙏🙏
Hi
bro can you send me the material
Just started watching your videos. You explain the concepts in a simple manner.Thanks
thank you so much respected sir. Alot love for you from pakistan. this video was very helpfull. we are looking foreword to see others playlist like these from you. once again thanks
Hi Krish, It's the best video I have ever seen. Crystal clear.
Still the best video out there. I think other content dont know what a practitioner of DS needs at 2:30 am .... :p
No Words for education. Many Thanks and wishes for futures.
We need mentor like you... Great job👍
thanks a lot, this thing can't be explained better than how you explained it.
I just became Fan of your ML knowledge.
Hi
Krish your way of explanation is just amazing....Thanks for these amazing videos and yes please share zip file
Hi
Hi
Wao thank you soo much, sir you explained soo well. whenever I face any doubts your video saves my day.. God bless u .. Happy Learning
you are doing a wonderful job Kris...👏👏
thanks sir for listening to my request to create a video on mean encoding , i am really enjoying your videos , and i have learned a lot from that. Please continue to create such awesome videos.
Hi
Sir, please share the link once again. I saw your video and it's a very helpful for the student's like me. I want to know more about the feature engineering.
Thank you for making such an amazing lecture. Waiting for the feature engineering link.
Hi
Krish Sir the way you explain is easy to understand. Please reopen the form. Thanks 🙂
Thank you for putting the time and efforts to create this video, also all other videos. Very helpful.!
Hey krish, nice video as usual... Filled the form and thanks for making motivational and additional support videos for encouragement. Kudos
Hi bro, could you please send me featuring document pls?
@@SanthoshKumar-dk8vs you can fork it from either mine or krish's GitHub account. Check Krish's video description for his GitHub link and you find all there
Hello bro, can you share zip file, bcz I watched it today so not able to fill form as you know.
Kaushalshivam2018@gmail.com
hi bro this is sarath..
I am a data scientist aspirant can you share me feature engineering notes..
mail id : sarath20994@gmail.com
Hi krish, nice way to collect the data free of cost.
you saved my day with mean encoding
Sir u r doing really great and I think under your guidance I will become a good data scientist soon...please help me sir
I liked the Mean Encoding technique and Target-guided encoding. We are preserving the normality of the data as well as not increasing the dimensions.
Thanks Krish Bhai..I have learned a lot from your videos
Hi Krish, it is just a suggestion if u start same channel in Hindi language. It will more helpful to those Indian students who are living in small cities and not much familiar with English lecture. Hope u understand my request. I'm your regular viewer and respect ur effort and knowledge. God luck.
Amazing explanation sir 🙏🙏
Hello sir...your way of teaching is really incredible.
I am studying through your lecture for past 1week and that's why unable to fill the form to get the materials which you have prepared for the same...
So if possible please enable the form link again...
Thanks man! Great content The Lord bless you with more understanding and help you to know Him better and better
Please re-open the form for feature engineering techniques. Thank you.
Yes sir please re-open the form
Excited to learn the coding part too Sir.
Great help Krish... Thanks for your video man
Vishal Shukla. could you please share this docs with me on dolly.shukla7860@gmail.com
Hi
Very useful information provided by u sir. Thank you.
You are the best sir.
so happy I found your channel...wooh amazing lecture
Please send me the zip file with respect to feature engineering
thank you sir
will definitely join your channel.
Thank you so much for sharing your knowledge with us
The video is quite informative and easy to understand. I really loved the video :)
I came across this video today and i like to learn more feature engineering
"if you don't mind will u reopen the link sir"
Yes pleaseeee
Yes please sir reopen
Yes, it's very much needed now
yes please sir
It's on the GitHub
excellent job Boss. really helpful
Excellent Explanation Sir, Thanks a lot
Guys plz if u don't like his videos then leave it, but don't do dislike 🙏
Hi @Krish, can you please share the Feature engineering materials if possible. Your videos are really impressive.
Thanks sir for all these free contents! :p
Can you send the zip file to me.
arifmollick8578@gmail.com
Thank U so much Sir for such Huge help....
Really good one Krish
@16:18 position you are saying to use 'one-hot encoding with multi-category' for an ensemble technique. But the beginning of the video you had explained ensemble techniques does not require feature scaling. Can you please clarify?
Thanks alot for sharing such a absolutely amazing knowledgeable video...
Thank You Sir!!things we can understand easily by your Videos.Sir could you pleasee reopen the link where we could get the Feature engg materials that could be more great
Clearly Explained, Thankyou!
Hi Krish , I really liked the way you are teaching, could you please share the feature engineering study material?
Great Video
plz give demo also
Hi krish,
I started seeing your videos now and want the feature engg doc. Can you please open the link for the form?
Waiting for your response.
Sir , Thankyou for this wonderful lecture , please share the study material
You are awesome sir 🙏
Hi
Nice information about feature engineering. Thanks a lot
can you plz send it to me
In mean encoding,If the feature values are replaced by the mean values ,the no of data values in the pincode column are still the same right??Then whats the point of doing mean encoding?
thank you sir from tamil
Hi Krish,
At 20:23 the Label for A - 0 and A - 1 will be different based on mean right ?
for example the mean will be calculated this way right ?
A - 1 => 0.73
B - 1 => 0.6
C - 1 => 0.4
A - 0 => 0.5
B - 0 => 0.35
C - 0 => 0.36
Then the ordering of feature will be as below right ?
A - 1 >> B - 1 >> A - 0 >> C -1 >> C - 0 >> B - 0
Hi krish, thanks so much for shedding light on this topic of Feature Engineering. I'm at Beginner Level of learning DS/ML and I really fell in love with your way of teaching these techniques. I would really love to get that document on FE you mentioned about in this video. I tried to drop my details via the google form but I see it's closed. Kindly assist please. Thanks in advance!
Sir, I am working in Data Science for a long time but want to your all playlist as I already have covered some of them. I need your notes on Feature Engineering so can you provide me it now. I shall be very thankful to you for this kindness.
Best wishes more love for you from Pakistan.
Hi
Krish naik you have been so helpful to me on this journey to datascience. The form is no more available is there another way for me to get the feature engineering and again about the Patreon are you still running it or it's already closed?
Hi krish.. google response link not active. how can I get the material
Great Video, Thanks!
Hi Krish, I have started liking your channel so much. Hats off for the great service you are doing for the aspiring and already experienced Datascientists. The form url which you have shared is no more available. Could yuo please share the material via google drive or reactivate the form.
by introducing a higher number to the categories on the basis of a higher no. of occurrence in a given class ( say here 1) are you not introducing bias in the dataset? ( target guided ordinal encoding)
Thanks Krish!!!! :)
Hi. Happy New Year Krish. Your content is awesome.
Can you pls reopen the link to the encoding document?
Hi sir,
I want the feature engineering doc. Can you please open the link for the form?
Waiting for your response
Hello I want the feature engineering document👏👋👋👋👋👋 Just came across this video please
@@MM-vx8go its available on his github
Akhil Kasare where please
Akhil Kasare this is my email.. mmaxwell265@gmail.com
What's the github username
Hi
I already joined as a member
Thanks for sharing, the video is helpful!
@krish Naik I have a doubt. FOr mean encoding and target guided encoding we need labels for encoding but how would we encoded the data at test time. ?
Hi Krish ..... I had recently started to follow up your video & it was very helpful. could you please provide me the materials related to feature Engineering......thanks in advance
31 dislike for what? Teaching you free of cost with market standard!!
One should provide the link of better videos,if they dislike anything. 🥇
Hi Krish, I have a question.
So, we are normalizing all numerical fields which will make it in the range 0 to 1.
While encoding categorical variables, we "one hot encode" a few which will make them either 0 or 1.
However, there are a few categorical fields where we are "label encoding" as order matters. It would be encoded like 1,2,3,4,.... which is not in the 0 to 1 range unlike the first and second case that I had discussed. I would like your advice here whether I need to normalize after label encoding the data and range of this case alone won't be in 0 to 1 and normalizing would make it in 0 to 1 range.
Hi Krish, I am confused with your explanation. My doubts are: You said for target guided ordinal categories you are assigning the rank based on mean values then how does it matter if the category is nominal or ordinal since the ranks are assigned based on mean values and not the inherent rank/order of the variable itself. Also for label encoding the number actually mean anything since it isn't like price/sales where the number holds significance so won't the result be junk/unusable and if they are usable then how do you interpret the result
If the categorical feature is ordinal then we can assign labels to it. Here the category which obtains highest target mean will be assigned the highest label value. If the categorical feature is nominal then we cannot assign labels.
This is because in label encoding the number does mean something. Different numbers teach the model to make different predictions. For example, a value of 4 (PHD) in the salary prediction example results in a prediction of higher salary whereas a value of 1 (Bcom) results in prediction of a lower salary! This happens because, in the training data, entries with PHD as educational attribute will have a higher salary in the target column. This is generally useful when we do not know the exact ranking!! We find the correlation on the categorical feature to the target and then rank categories according to the mean of the values observed in the target!
If the categorical feature is nominal, we do not want the algorithm to learn more from some categories compared to the others. Hence using one hot encoding, we set the values to be 1 and 0.
Now as for Mean encoding of nominal categorical features, we essentially map a relationship between the categories and the target. When applied to the example of pin code numbers, some pin codes may result in higher salaries (assuming we are trying to predict salary again). Hence the mean of that pin code will be higher. We simply map in the mean value in place of that category!! So now when the model learns, it will know that a data point with high value in the pin code feature should predict higher salary!
Regards!
Hi @Krish
Bro Your link for google form is no longer available.
And I must thank you for the amazing knowledge you share with us. God bless you for your good work. Trust me, I think you are the only person who can teach data science to anyone, I mean you are making it halwa for your viewers :D. I have watched almost all of your Videos. Thumbs up to you (y) .
Hi
Amazing and easy to follow explanations! Newly subscribed and loving it! Just curious how you recover the particular categories and make sense of your results if you use something like mean encoding. Do you have to trace back the original definitions for each mean and what happens if there are repetitions?
Hi
Please do a session work on the dython package and setting categories in it
Hi sir.
I have started to learn ML from your channel only. Thank you for your knowledge that you are sharing with us.
I also have one request for you can i get feature engineering zip file now. I am really interested in ML.
Thanks a lot for the clear explanation. Can you Please reopen the google form again?
Hey Krish! I have few questions based on encoding which are
1. Let’s suppose I have a feature which has 1000 different categories which I need to convert to either an integer/float how should I do that?Here I can’t go for one hot encoding as it might create 999 columns.And also it has only 1100 record /rows by which even though going by the “one hot encoding with multiple categories” method the most repeated categories will be extremely less how do we handle it in such cases?
2. In Ordinal encoding why are the ranks need to be assigned to a categorical label instead we can give some random unique number to the categorical values without ranking them for example PhD as 1 , BE as 2, Masters as 4 and Stats as 3.
3. Also regarding “Label Encoding” how are the ranks decided say if PhD needs to be given higher rank let’s suppose ‘4’ how can a library know that it should be given a higher rank? Or is it something else that in the library code we need to manually set it?
Do you have any ideas how to tackle this issue ?
For the first point u should not apply one hot encoding instead we can go ahead with Mean encoding.
Label encoding for ordinal categorical will be assigned with ranks. In this case PhD should have a highest rank or label. This will help us to specify the ML algorithm where in we are providing higher importance to phd
Krish Naik Thanks a lot Krish👍🏻😊 And thanks a ton for your awesome content learning new things. Waiting for the part - 2 of the series😊
Hi sir,
I need the feature engineering material sir.so please send the link
Once again..your techings are awesome and clean to understand 🙂
Hi @krish Naik, how can get the zip file of all feature engineering techniques? kindly help
Please share with me too. Thanks
Could you please upload the forum again . ?
Thanks in advance :)
Sir please open the form enteries to get zip file for feature engineering
Hey Krish, can you please do share the ZIP file which you have mentioned in the video about the Feature engineering, as I am unable to open the Google url link. it will be more helpful if you help me with the file.
Sorry for the stupid question: what´s the output? I mean you encode before you apply any ML algorithm and to that point you just have the dataset, what kind of output do you mean here? Thanx
I agree with all the encoding you treated, however I got doubts on assigning an ordinal label based on the mean value of the output. By doing this you are setting the distance between levels to one, while in reality, based on the output mean, some level might be close than others, with distances less than or more than 1. What do you think?
I would also point out the risk of data leak in case of mean encoding, for this reason I'd add a random noise to the mean
Hi krish,
Today only I saw this video.could you please kindly open that Google forum one more time.
In Target guided encoding if mean of two variables are same then how to assign numbers? As both has same mean how to decide for which has to give more numbers.?
This is very informational video. I would like to go through the feature engineering material mentioned here. Will it be possible for this material to be available?
Sir I have a dataset 5000 rows and 36 columns. It is a mixture of both string and interger values. There are are two columns Name and Company which should be encoded. Which method would be the best?
Thank you so much Kris, your videos have been instrumental in my learning process. However, I wanted to find out if it were too late to get the feature engineering code. As it mentioned, the link was opened fir two days hence its been over a year. Or if you have it on your git, I can get it too. I've been struggling with feature engr, though I'm still learning, having that code would mean alot to me. Thank you so much in advance
Hi Krish, I am not able to fill the form. Its removed. Can you please upload that
Same here
same here
Where did krish upload the form... Can u share the link related to it
@@sravanijammula573 Krish uploaded the form when he uploaded the video. Now it's old so I think he removed that. I am also not able to fill the form as I saw video very late
Thanks vishal for the update... If u are aware of it jus post it here...
1st to view, 2nd to like, 1st to comment.
Hi Krish, thanks for this very good video. Just to clarify a bit, when you say Pincode, are you referring to the postal or zipcode? Because from where I am, when they say pincode it means bank's pass code or something like that.
in one hot encoding for mult. category if top 20 category has same count then what should be done
Hi sir how is is going to be of target guided encoded and mean encoding in case of regression problem.?
How would we calculate the mean if the output is a multi class classification. In that case shall we take 0,1,2 as output?
In the eg you have taken 0&1. Here we can do the calculation. What if there are more than 2 classification outputs.
If you could attach the notebooks in a link it would be easier i guess instead of sending personal emails. Just a suggestion.
Hi
Sir please could you please tell us why the theory of computation is actually used and what are the application of these subjects please Sir make a video on that
Hi everyone , does deleting one dummy variable column is automatically done by onehotcoding ? Or it should be done mannually
Hi Krish,
We cannot perform Mean or Target encoding on test data because we don't have target column in test data. So how can we deal with such a situation where we have variable with multiple level in it?
I am talking in respect with Hackathon where we generally don't have target variable, this is something which we have to predict.
Would appreciate your help.
You already got the ordinal number or the float number for each category class from the training data . So you dont need to do it again in test data. You will simply use it.
You might already know this.
But I am answering if someone else has this doubt.
It's very helpful, sir please reopen the form link...