Complete Exploratory Data Analysis And Feature Engineering In 3 Hours| Krish Naik

[@krishnaik06]

Give this video 1000 likes then I will start a 7 days Live NLP community Sessions for everyone. Happy Learning!!

[@vinayakdumbre2828]

Wow,its should be end to end,not just basic rnn,it would be awesome

[@photogenicglint239]

Hi Krish , Collab with Sumit Mittal ( Trendytech) for big data course. He teaches in depth but offer course at high price.once he Collab with ineuron so that he can offer course in affordable price.

[@vivekpandey8438]

thanks please start NLP common file and Also Upload statistics in 1 Videos

[@aryansheth7369]

666

[@faraazmohammed3693]

992..close

[@mainlykanchan8740]

Sir data analysis in sql with advance queries for portfolio project. Full length video, like this video please 🙏🏼

[@krishan9739]

I have to say, this actual teachss you how to think for EDA so you actually learn something.

[@najiibrashiidabdi5014]

My name is najiib and i from country called somaliland which is in somalia really i enjoyed this project i will hope you will upload more topics about machine learn thank you krish naik najiib from somaliland

[@mayowaolowolaiyemo1606]

Thanks for this teaching Krish, your approach is simple and easy.

[@thedinaaesh]

01:30:08 - Based on my experience, when we encounter null values, we typically reach out to the upstream data source to verify if there are any missing values. In most data science projects, data is provided by a data engineering team, so collaboration with them is essential. If they confirm that the data indeed contains null values after their validation, we can then handle those nulls accordingly.

[@VyomKumaraes]

47:50 this is also working df[df['Aggregate rating'] == 0]['Country'].unique()

[@rishi4307]

# Function to convert duration to minutes def convert_to_minutes(Duration): hours = 0 minutes = 0 Duration = str(Duration) # Ensure the duration is treated as a string if 'h' in Duration: hours = int(Duration.split('h')[0]) Duration = Duration.split('h')[1] if 'm' in Duration: minutes = int(Duration.split('m')[0]) return hours * 60 + minutes # Apply the function to the 'Duration' column final_df['duration_minutes'] = final_df['Duration'].apply(convert_to_minutes) final_df.head()

[@adityagoyal6527]

At 47:54 the same can be done using final_df.Country[final_df['Aggregate rating']==0].value_counts()

[@shashwatgoswami6994]

Very informative video. I would like to add a point regarding the UTF-8 code error i.e if you save the excel sheet as CSV UTF-8 comma delimited format then there is no need to enter the codes.

[@loserianlaizer4945]

thanks Krish..it has been an enlighten session.. Have watched the entire 2.48hours session. Be blessed

[@aishwaryapattnaik3082]

Label Encoder should be used only for target labels i.e y and not on input feature. It's mentioned in sklearn Label Encoder page clearly. For nominal & ordinal variables, we should use One Hot Encoder and Ordinal Encoder respectively. These all should be done within a pipeline and column transformer for hassle free coding preferably

[@rajkundra5005]

yes,same doubt

[@prayashdash1815]

@@rajkundra5005 bhai link dede

[@NikhilSingh-gv5ne]

Mind-blowing explanation bro keep it up

[@Agros92]

Thanks Krish, you are the best!. A question related to the "second session" about the Product_Category_(1,2,3), I understand that you explain that in case of NaN values in categorical feature you can use the Mode to replace the NaN values. But for this particular case I think that is important to understand the data before doing that, since Product_Category_(1,2,3) indicated that the products can be part of multiples categories. For example a movie being categorized as "Drama, Action, Suspense". So for this case maybe it would be better to try to use dummies for Product_Category_(1,2,3) and then try to sum it, it would be complex to implement it but you would get the real information about your data, since you can get the info about Product_1 being a (0,1,0,0,1,0,1) if that product has 3 categories. Cheers!

[@photogenicglint239]

Hi Krish , Collab with Sumit Mittal ( Trendytech) for big data course. He teaches in depth but offer course at high price.once he Collab with ineuron so that he can offer course in affordable price.

[@celebrationsthecelebschoic575]

When you try to get top 3 countries percentage in pie chart, it calculates for only those three countries. But calculating over all the transactions will make sence. Percent of transactions from India means, among all the transactions what is India's percentage. But here in hour case, it allows only India, USA and UK.

[@dikshantakumarbharadwaj6052]

God-Father of Data-Science

[@raosajid6578]

great work sir subscription done from my side

[@shivamkumar-rn2ve]

There are two types of variable nominal and ordinal In ordinal you can use label encoding but you can't use label encoding for nominal variable you have to use one hot encoding if you will use label encoding for nominal then machine learning model will treat nominal as ordinal so you can't use

[@aishwaryapattnaik3082]

Label Encoder should be used only for target labels i.e y and not on input feature. It's mentioned in sklearn Label Encoder page clearly. For nominal & ordinal variables, we should use One Hot Encoder and Ordinal Encoder respectively.

[@shivamkumar-rn2ve]

yeah you are right about label encoder you can only use it for target variable

[@learner8053]

Please post EDA video in your hindi channel also

[@arbiiimesh]

Hi, Sir I'm an aspiring Data Analyst enthusiast and IT assistant.. Now become a member of your channel.. 🙏 please guide me what needs to be learnt first for a data analyst or data science career.. Hope you will reply Best Regards

[@HiralPrajapati-j5x]

Query for flight price prediction dataset for duration column df['Duration_hour']=df['Duration'].str.split('h').str[0].str.split('m').str[0] df['Duration_hour']= df['Duration_hour'].astype(int) It's work for me.

[@yogeshmane9973]

you are doing excellent work sir

[@praveentanikella4078]

Nice one. One doubt the main work of data analyst is only finding insights and done. The ML part no needed?? Is that ML job work is for Data scientist.

[@yumatinikhar7858]

Thanks. Its really helpful

[@swapnilpalsapure9781]

Really helpful Sir..

[@kar2194]

Hi Krish, do you have videos of data cleaning, EDA, and feature engineering for unsupervised ML? (For both Principal Component Analysis (PCA, CA, MCA... etc) and Clustering techniques include partitioning, hierarchical, DBSCAN etc). By the way, are there differences in cleaning cleaning and feature engineering between predictive regression and inferential regression? Thank you!

[@knowledgedoctor3849]

Great Sir❣️

[@ramdasprajapati7884]

Find the top 10 cuisines(food) item for this for zomato dataset is this code correct final_df.Cuisines[:10].value_counts()

[@vijayramapple]

53:15 / 2:48:54

[@ninadkalanke8880]

Sir plz turn off your notification sound!

[@SACHINKUMAR-px8kq]

thank you so much sir

[@adamassrkfan]

2:36:35

[@SachinModi9]

How to find top 10 Cuisines final_df= final_df.replace(np.nan,'Dummy') --- Convert NaN to Dummy one_string = ','.join(final_df['Cuisines'].tolist()) -- Convert Cuisines columns to list and join one_list = one_string.replace(" ","").split(',') -- replace blank spaces by comma pd.value_counts(one_list)[:10] --- top 10 values

[@MujahidDeen-x6o]

Make another video in data explratoery, eda

[@vikasvs5755]

super

[@Rijuldhungana]

data[data['Aggregate rating']==0]['Country'].value_counts() , This also works

[@pratik5692]

feature engineering in one video

[@ayushsharma5640]

Thanks sir

[@yasmeenkarachiwala9612]

Hello Sir! Thank you. @43.00 why the observation of the maximum number of ratings is from 2.5 - 3.4?

[@GuruprasadP-s6s]

We could have used product ID to fill product category column

[@Amansharma-he9qg]

first comment sir how to make sql project for portfolio please reply

[@mainlykanchan8740]

Yes..

[@praveentanikella4078]

For data analyst work the data set is available from any data base or in form of excel or CSV ??

[@siddhantgaurav7053]

feature engineering in 1 video

[@narayanbabubharali9846]

Nice

[@usmanriaz6157]

Sir, Airline is a nominal feature and in you said that in case of nominal feature, we can do OHE or Mean encoding. Why are you using LabelEncoding ?

[@RajaSirOpsc]

where is the blackfriday dataset

[@sonukumar-yp6vs]

11:00

[@dibyanikshetry3775]

I couldn't do the part where we have to show the country names that has given 0 rating It's not showing any output

[@aakashpal0777]

Hi

[@sangramshinde9262]

I dont understand replacing na values of product catogry_2 and product catogry_3 with mode we just manipulated the data

[@jagadeeshct7083]

please share blackfriday dataset ..there is no blackfriday dataset in the given link.

[@kirankumar9934]

Even I'm not able to find black_friday dataset

[@bestofmusicc__]

Hi, why did you combined the country code ?? Please explain this.

[@A3dull]

The first dataset only includes the country code, while the second dataset contains both the country code and the country name. When merging them together, the country name column was populated using the information from the second dataset.

[@bestofmusicc__]

@@A3dull yeah thanks man👍💪

[@jececdept.9548]

is this a regression problem?

[@thedinaaesh]

01:41:40 - probably few men are buying on behalf of the women 😂

[@garimabatra2658]

how to download dataset? pls help

[@elahehkhazaei4855]

Please give us source of you data

[@anonymous_12155]

I am getting Nan error when I try to replace F with 0 and M with 1 in Black Firday EDA ..How to resolve it?

[@pavankumarjammala9262]

Once before running that particular code run all cells at a time you will get it

[@vijaysharma7677]

please explain how one can find the location of CSV or get the jupyter NB to read the file location automatically inside a folder I am getting an error while reading the file

[@ManishKumar-qh2ql]

open with path location and instead of \ use \\

[@madhupincha7898]

pwd()

[@Agros92]

You can put the csv file on the same folder of the JupyterNB file. To read it it would be - pd.read_csv("data_name.csv") -. If you put the data in another folder and that folder is located in the same folder of the JupyterNB file you can do - pd.read_csv("Folder_Name\\data_name.csv") -

[@ShikhaJain-u7y]

I am not finding train.csv file for the second part of video in your github

[@himadrikar4664]

The link to the data set is given in the first line of the Python notebook. Download the dataset from that link.

[@srirama8275]

What are Prequesties to learn this sir?

[@krishnaik06]

python

[@srirama8275]

@@krishnaik06 Thank you sir

[@abhisheksinghmahra446]

sir how to deal with utf-8 encoding

[@Srushti_Mane]

use latin=1

[@azamiqbal8792]

Can you share file for practice

[@tanumoyhazra6055]

from where i can get your codes for this video ?

[@vanshsrivastava6551]

Is this enough to mention in resume

[@prafulaggarwal9683]

where to find the black friday dataset?

[@swetamishra3580]

Did you find it?

[@himadrikar4664]

The link to the data set is given in the first line of the Python notebook. Download the dataset from that link.

[@oseikofi4953]

I can't find the black friday dataset on your github page

[@himadrikar4664]

The link to the data set is given in the first line of the Python notebook. Download the dataset from that link.

[@moghalkarishma2378]

Is necessary to hanle missing values in data analysis?

[@_k_kd]

yes.

[@RolandElvira-l4y]

Wilson Larry Moore Richard Martin Daniel

[@Pyrometin]

Guys how to find top 10 Cuisines in data ? help me

[@Pyrometin]

I got it, use this code. final["Cuisines"].value_counts()[:10]

[@Abhi-qn4xv]

Can anyone explain when do we use onehotencoding and when do we use Labelencoder(ordinal encoding) since they both do the same job but in a different way, onehot creates multipe new feature while label do all the work in one feature. Like in this case wouldn't be better to use labelencoder to do encoding in Additional info feature since onhot will create multiple new sparse eatures which might increase he workload of the mode or am i missing some point here?

[@sanjaysanjay862]

One-hor encoding is used only for independent variables (feature) but label encoder is used for target variable.And they both won't do the same task one-hot encoding gives seperate columns for each catagory.As of my understanding.If wrong reply

[@Abhi-qn4xv]

@@sanjaysanjay862 well u r correct. I did some reading in this topic and found out that although label encoder can be used on independent variables too, it's usually not used. On independent variable, one hot is better than label encoder as label encoder might confuse the model into learning that feature as a rank. So instead of learning 1 as a numerical representation of a word, model will think 1 as a rank. Hope u understand my point

[@sanjaysanjay862]

@@Abhi-qn4xv Yes, I agree that

[@adeshinaibrahim9641]

In simple terms use one-hot encoding when you have limited number of categories but otherwise dont.

[@muhammadzakiahmad8069]

Zomato Dataset Assignment: (With respect to value counts) cus_values = final_df["Cuisines"].value_counts().values cus_labels = final_df["Cuisines"].value_counts().index plt.pie(cus_values[:10],labels=cus_labels[:10],autopct='%1.2f%%') (With respect to Aggregate rating) final_df[['Aggregate rating','Cuisines']].groupby(['Aggregate rating','Cuisines']).size().reset_index().tail(10) Please correct me if i did it wrong.

[@FaizanSharif-k8g]

final_df= final_df.replace(np.nan,'Dummy') --- Convert NaN to Dummy one_string = ','.join(final_df['Cuisines'].tolist()) -- Convert Cuisines columns to list and join one_list = one_string.replace(" ","").split(',') -- replace blank spaces by comma pd.value_counts(one_list)[:10] --- top 10 values

[@PradeepSahu-kh8vr]

im not getting zomato csv file....can anyone help????

[@pavankumarjammala9262]

Yeah !! bro same prblm from my side also

[@ashishsaha6904]

why latin-1 ?

[@naveenojha8377]

Hindi m hota to jarur kuch Sikh pate 😓😓😓😓

[@jackymarcel4108]

Lewis Barbara Davis Brenda Brown Michael

[@prashantgupta2172]

Hindi me vedio bana digite aap

[@NooBGamer-fd4ln]

he said fucked instead of fixed 1:51:00 😆

[@jedits7835]

after do doinh this project can we add this resume

[@MLMinute]

Everything is perfect except the pronunciation. Haha

[@UniversalFacts-unknown]

How to give just zomato.csv in df line instead of giving entire path

[@himanshutola3729]

Keep the CSV and the ipynb file on same folder