This was very very informational. I was recently interviewed for a data science role and i was also asked questions on cnn and transformer model. I answered everything related to cnn but i had no idea on nlp transformer model. It has helped me a lot by watching this interview. I was also asked on the recent popular models like gpt2 and gpt3. I would like to thank Sudhanshu sir and Krish sir for giving their time. Thank you.
Sir just one small request, whatever questions u guys ask in the interviews pls try to pen down it and later try to provide answer of it in the description or later pin down the comment.
@krish sadhanasu explanation with CNN is partially correct .....we can build time-sequence CNN+LSTM to do video analysis. Only reason CNN is used to extract features and reduce features but here we already have the feature . So we can directly use LSTM > RNN . Also we had done one mistake we have aggregated the data on year and removed month. We should have aggregated to a month and then based on EDA we could use SRIMAX or ARIMA
I am Fresher and I don't know python but I am planning to persue my career as data analyst from last 20 day's so my friend suggested me to watch video from your UA-cam channel. really it is wonderful interview. please sir keep it up 🙏🙏🙏
55:10 The biggest disadvantage of LSTM/GRU is they're incapable of Transfer Learning, No support for parallel distributed training, 'tanh' activation functions and also vanishing gradients
Yup.. I've also experienced vanishing gradients with LSTM timeseries. On the contrary , I was able to generate decent forecast with CNN (Conv1D) using leakyRELU.
Good interview. I watched the whole thing. It would be helpful if you could tell if the candidate would be hired or not for a data science role based on the performance in the interview...
8:59 python question : I've a question. Why would I apply a filter to select only those records from the dataset that have floating point values in day column whose datatype is already "floating point"? To select records containing string values in Gender column whose datatype is object we have to apply filter because object type column may contain numeric values enclosed in quotes. Note: I have not watched complete video yet.
21:24 the solution presented selects entire columns of the dataset whose datatype are float and object; it seems the question asked was to filter the"data points" a.k.a records which have float values in day column and string values in Gender column. Kindly clarify.
After watched this video, I realized that now I am able to crack any Interview related to NLP and CNN. How can I approach having a Virtual Mock Interview on Computer Vision using Deep Learning and in General Data Science?
Q1 soln: df['day_temp']=df.day.apply(lambda x: True if isinstance(x, float) and np.isnan(x)==0 else False) df['gender_temp']=df.gender.apply(lambda x: True if isinstance(x, np.object) else False) df[(df.day_temp==True) & (df.gender_temp==True)].index OR just df.dropna().index Q2. sorted(s1.lower())==sorted(s2.lower()) Q3. df2 = df[['year', 'gender', 'births']].groupby(['year', 'gender']).sum() df2.head() will get table having male and female births year wise which can be scatter plotted.
i was confident and applying for data science fresher role. i was frustrated,why i am not getting interview call. i thought i can easily ace teh interview , all i want is interview call. i dont know why this video came to my feed. after struggling 20 minutes i got thisss... df[["gender","day"]][df["day"].notna()] after this i am brokken.. i took 20 minutes to solve this simple thing 🤧🤧🤧🤧🤧🤧
Hi Krish/All, can someone provide me any video link or any resource where I can practice these kind of practical questions like the one asked by the interviewee (python questions) just so that I can brush up my skills.
mylist = list(df.select_dtypes(include=['float','object']).columns) is this right can someone tell me for the first question if we don't use .columns we will get true false which col don't have the datatype in day or gender. Please tell me if this is the solution if some can.
Is anybody trying to find data science related jobs in delhi/ncr cause i am to ,but there aren't any ..most of the data science jobs qre in Bengaluru pune Hyderabad
This was very very informational. I was recently interviewed for a data science role and i was also asked questions on cnn and transformer model. I answered everything related to cnn but i had no idea on nlp transformer model. It has helped me a lot by watching this interview. I was also asked on the recent popular models like gpt2 and gpt3. I would like to thank Sudhanshu sir and Krish sir for giving their time. Thank you.
hi, pls can u tell me what is the main courses to be a DS? thx im a bachlor degree
@@hideweapon1361 pls follow krish sir's youtube channel. he has given the whole roadmap to how to become a ds.
Sir just one small request, whatever questions u guys ask in the interviews pls try to pen down it and later try to provide answer of it in the description or later pin down the comment.
@krish sadhanasu explanation with CNN is partially correct .....we can build time-sequence CNN+LSTM to do video analysis. Only reason CNN is used to extract features and reduce features but here we already have the feature . So we can directly use LSTM > RNN . Also we had done one mistake we have aggregated the data on year and removed month. We should have aggregated to a month and then based on EDA we could use SRIMAX or ARIMA
I am Fresher and I don't know python but I am planning to persue my career as data analyst from last 20 day's so my friend suggested me to watch video from your UA-cam channel. really it is wonderful interview. please sir keep it up 🙏🙏🙏
@ZEEME AHMAD thank you
This guy interview is like my college day's viva..where I used to explain all histories except answer.
hahahahahahaha
Truee
55:10 The biggest disadvantage of LSTM/GRU is they're incapable of Transfer Learning, No support for parallel distributed training, 'tanh' activation functions and also vanishing gradients
Yup.. I've also experienced vanishing gradients with LSTM timeseries. On the contrary , I was able to generate decent forecast with CNN (Conv1D) using leakyRELU.
Lovely concept sir , got to learn a lot. Thank you all of you.
Amazing discussion..a lot to learn from this..Thanks for sharing it with open platform..🙏
Good interview. I watched the whole thing. It would be helpful if you could tell if the candidate would be hired or not for a data science role based on the performance in the interview...
you can't say confidently as there are not other interviewees to choose from
Brilliant learning... Thanks for this level of content.
Check for seasonality and trend use either additive or a multiplicative aproach
8:59 python question : I've a question. Why would I apply a filter to select only those records from the dataset that have floating point values in day column whose datatype is already "floating point"?
To select records containing string values in Gender column whose datatype is object we have to apply filter because object type column may contain numeric values enclosed in quotes.
Note: I have not watched complete video yet.
21:24 the solution presented selects entire columns of the dataset whose datatype are float and object; it seems the question asked was to filter the"data points" a.k.a records which have float values in day column and string values in Gender column. Kindly clarify.
Sudhanshu is superb while explaining the concepts...
Could we have more videos on projects and concepts rather than virtual interviews?
After watched this video, I realized that now I am able to crack any Interview related to NLP and CNN.
How can I approach having a Virtual Mock Interview on Computer Vision using Deep Learning and in General Data Science?
Great please keep on do this job
Thanks
Love this!
Q1 soln:
df['day_temp']=df.day.apply(lambda x: True if isinstance(x, float) and np.isnan(x)==0 else False)
df['gender_temp']=df.gender.apply(lambda x: True if isinstance(x, np.object) else False)
df[(df.day_temp==True) & (df.gender_temp==True)].index
OR just df.dropna().index
Q2.
sorted(s1.lower())==sorted(s2.lower())
Q3.
df2 = df[['year', 'gender', 'births']].groupby(['year', 'gender']).sum()
df2.head()
will get table having male and female births year wise which can be scatter plotted.
I'm not so fluent in English.. Should I worry?
Same me too
No, but just be able to communicate what you are typing
Not at all
Yes
Check with isinstance()
i was confident and applying for data science fresher role. i was frustrated,why i am not getting interview call.
i thought i can easily ace teh interview , all i want is interview call.
i dont know why this video came to my feed.
after struggling 20 minutes i got thisss...
df[["gender","day"]][df["day"].notna()]
after this i am brokken.. i took 20 minutes to solve this simple thing 🤧🤧🤧🤧🤧🤧
Bro at least you did something. I didn't even get it. Btw what happened to ur job process?
he just reading the answer...when they talk he looks upward and when asked qauestion he saw on screen
Hi Krish/All,
can someone provide me any video link or any resource where I can practice these kind of practical questions like the one asked by the interviewee (python questions) just so that I can brush up my skills.
Ans for Q1-
df[(df['day']=='float64') | (df['gender'].dtype=='O')][['day','gender']].dropna()
Nice
mylist = list(df.select_dtypes(include=['float','object']).columns) is this right can someone tell me for the first question if we don't use .columns we will get true false which col don't have the datatype in day or gender. Please tell me if this is the solution if some can.
and also dropna() in the end
Notification ni aaya iska
videos say Mock Interview For Data Science Role but most of the questions were deep learning-based.
Deep learning work is also done by data scientist
Whenever I see krish sir...he seems to look like ashish chanchalani
Is it too late to change the career at 32 in data science?
Yes
Its never late.. you should have determination
You can easily transform your profile
@@gunnu61lol who are you to decide what age is right? I shifted at 32 too. So maybe stop demotivating others.
nice
Is anybody trying to find data science related jobs in delhi/ncr cause i am to ,but there aren't any ..most of the data science jobs qre in Bengaluru pune Hyderabad
Real interviews arent much like this