Kaggle - Titanic Solution [1/3] - data analysis

[@MadhuKumarKilli]

Wonderful step-by-step data analysis of a complicated dataset. Thanks for your video. Was very helpful in guiding me through the initial phase of Data Analysis with Python.

[@anubhavsood1510]

Complicated? This is one of the easiest Dataset.

[@robertogarza1902]

This is a very good video and it helped me with an assignment for my machine learning class. Thank you very much!

[@ishaansharma5982]

Useful for getting started with kaggle competition . Nice.

[@TheEasyoung]

thanks for cheerful replying, I will keep up good youtubing :)

[@jeehoneybee2589]

Wonderful introduction on Kaggle. Thanks!

[@brianheartwood2071]

Good Video Minsuk, thanks my friend

[@mathError4004]

Excellent sir. Can you make a video about Advanced House Price Prediction problem from kaggle?

[@akshaymishra4957]

I was able to rectify it....Good tutorial.. Keep helping us with more data set.

[@bhavyar2986]

Beautiful explanation

[@prabirbiswas440]

Awesome. Thank you very much Minsuk. This is very very helpful.

[@creativecore3575]

Hello! Awesome video on this data set! It's very nice to see your way around the mystery as a Data Detective. But i have a question, how did you confirm on the SibSp data visualization section @ 9:15 that you were more likely to survive if you were accompanied by family members? The colors at the top of the [Dead] bar show that the people who came with many family members, no? Thanks again !

[@TheEasyoung]

Lannon Khau you are right. The chart shows more likely to survive if the guy have only few siblings, but many siblings more likely dead. Thanks for feedback!

[@kaushik4420]

Fantastic work , thanks a ton :)

[@Antnierv]

how is it that you are more likely to survive if you have more than 2 siblings/spouse? i see all the colors representing having more than 2 in the dead bar....

[@TheEasyoung]

Good catch and thanks! You are right. Having more than 2 actually died more.

[@Antnierv]

@@TheEasyoung okay thanks! great tutorial

[@vanshikapathak517]

What is the need to take title feature as we know we have sex amd pclass feature

[@dhananjaykansal8097]

You’re lovely sir. Beautiful explained

[@abrahamjacob8876]

Really informative video ! Thanks a lot !

[@russellmatson7923]

Great work! Would be better as a rap duo with a Titanic survivor or their descendents.

[@simsin3297]

Is it necessary to plot a graph..?

[@TheEasyoung]

Simran Singh nope! Just for better understanding of the data :)

[@simsin3297]

@@TheEasyoung thanks for replying.. I am a beginner in machine learning...so,how should I decide features for graph ?

[@TheEasyoung]

Simran Singh anything you think it is relevant, you plot it.

[@davidpage9681]

Fantastic video, I am trying to follow what you have done and am getting the folowing error when trying to apply the code from Binning 4.4.2 for dataset in train_test_data: dataset.loc[ dataset['Age'] 16) & (dataset['Age'] 26) & (dataset['Age'] 36) & (dataset['Age'] 62, 'Age'] = 4 ValueError: cannot set using a multi-index selection indexer with a different length than the value I have been on stack overflow and as I am still relatively new to coding am struggling ti undertsand this error, any help would be very much appreciated. Thanks Dave

[@bharatchandra2512]

NameError: name 'df' is not defined i'm getting an error that df is not defined 7:31 Please give fast reply sir How to solve the error?

[@anubhavsood1510]

You need to import the source data into the df first. Rather than coding, I would suggest going for theory and trying to see why are we doing what we are doinh

[@robind999]

very good video. thanks

[@stevesongrealestate]

I logged on Kaggle, but didn't see the button "my submissions' or "submit predictions ". Why?

[@TheEasyoung]

I logged in Kaggle and there is submit predictions button at the top right corner. I hope you find.

[@stevesongrealestate]

@@TheEasyoung I don't have on my side. Is there any requirement for the account?

[@TheEasyoung]

Steve Song I have no idea but at least I don’t remember if there was. I wish you find it soon!

[@stevesongrealestate]

@@TheEasyoung That's Okay.

[@manjunathnagendra9014]

Amazing. In plot function how to set default colors which you are using bar plot stacked. Because plot function doesn't have palettes, you can only set using colors

[@manjunathnagendra9014]

Hey I figured it out. sns.set() will do it. Thanks :)

[@Tino-df1su]

Hi! I watch your video about Titanic Solution today and I don't see your code in this link Github on the description. Can you share this code in video with me? Thank you!

[@akshaymishra4957]

my code is showing error when calling bar_chart('Pclass')..Help me out

[@TheEasyoung]

Akshay Mishra which error do you get?

[@satyammeena3155]

survived =train [train[survived]==1][feature].value_counts() please sir explain the use of feature in this and also explain the work of value_counts()

[@TheEasyoung]

in bar_chart function, here feature is given as parameter it can be 'Pclass', 'Embarked' or any feature. train [train[survived]==1] means I only choose survived train data, train [train[survived]==1][feature] means a list from the survived guys feature (if there are 100 survived guys, the list has 100 item in it in terms of the feature), . value_counts() will group 100 feature item into feature's class, for example, if feature were 'Pclass', value_counts() will give something like 1:30, 2:30, 3:40, since we have 1st class, 2nd class, 3rd class and assume we have 100 survived train data. I hope my answer helps your question.

[@vininitdgp]

Hi, Thanks for the code Everything is working fine but I got stuck at cell 77, error is showing at line 3 which is saying that "could not convert string to float: 'S' Pleaser let me know how to resolve this? Thanks in advance

[@caissa6187]

Thanks so much! 😊

[@joelkhaung]

Thanks for your demo. How did you generate Train and Test data csv. Any approach?

[@TheEasyoung]

Kyaw Khaung hi, you can get titanic dataset from Kaggle.

[@coricua88]

Great work, Prof. Heo. I have question for you. Is it indispensable or recommendable to alwasy divide dataset for train and test separately for beginners ?

[@TheEasyoung]

Yes. Since test data must be unseen data to the ML model for valid testing. Thanks for watching!

[@coricua88]

@@TheEasyoung so how can I select which datas to place into train and test dataset (csv file) respectively ? is there any rule to select for sake of accuracy ?

[@ChannelMath]

@@coricua88 they just need to be randomly selected

[@zeus1082]

But people who paid high fare or who departed from rich place is same as pclass.Why dont we just include jist the pclass and ignore others?

[@TheEasyoung]

you maybe right, you can test with your hypothesis, and compare the results to see which way results in better result in Kaggle test set. Sometimes we can delete duplicate feature or ignorable features. but sometimes data looks like correlated (when A goes high, B goes high) but which is actually not. thanks for good question!

[@zeus1082]

Minsuk Heo 허민석 I used forest algo and used sex,pclass,age,parch as features.I got 0.7655 score for that in kaggle.Just now I submitted.How bad is that score?.By the way the tutorial was superb

[@TheEasyoung]

aneesh cool your score is ok. Mine was 79.5 around. Cheers!

[@zeus1082]

Minsuk Heo 허민석 coooool.we could still improve our model right.Last time I took the median for the age.Iam going to include some features and see the result.Weould improve our score.

[@TheEasyoung]

good to hear that. one tip for you is ensemble. take a look bagging, boosting, stacking. since you already have preprocessed data, all you need is just collaborate multiple classifier results, which shall give you better results.

[@РадаСеношенко]

대단히 감사합니다 !!

[@deep1998-j1v]

Is this done in ML or DL?

[@TheEasyoung]

In ML.

[@pran6663]

thanks a lot !!

[@tamago9760]

参考になります！ありがとう。

[@pranjalpathak4498]

"It is the first class... very rich people" LOL

[@ChannelMath]

then what he finds next...not so funny

[@andrianjuric2732]

"Damn it Jin Yang"

[@4abdoulaye]

thanks

[@TheEasyoung]

Abdoulaye Diallo my pleasure:)

[@slawomirgontarek4213]

Please set translation on other languages. Thank for it!!!

[@guntherrondina2190]

Its so funny hearing him saying that most male are dead at 8:04

[@slawomirgontarek4213]

Hello, PLEASE SET TRANSLATION ON OTHER LANGUAGES !!! THANK'S

[@y_rb4080]

Man ur accent is insane :(

[@denisvoronov6571]

It's better to have it in English with an accent than to not have it at all.