Solving Real-World Data Science Interview Questions! (with Python Pandas)

Visit brilliant.org/KeithGalli/ to get started learning STEM for free, and the first 200 people will get 20% off their annual premium subscription
In this video we solve a series of Data Science Interview questions on Stratascratch. We start with easy problems using Python Pandas and then progressively get more difficult. At the end of the video we do five non-coding interview questions that force you to think at a high level.
Mentioned Resources!
Second Channel: / techtrekbykeithgalli
Regex Cheat Sheet: cheatography.com/davechild/ch...
Probability text book: www.amazon.com/dp/188652923X/...
Here are the questions that we complete (in order)
~~ Coding ~~
1. Finding Updated Records: platform.stratascratch.com/co...
2. Number of Bathrooms and Bedrooms: platform.stratascratch.com/co...
3. Counting Instances in Text: platform.stratascratch.com/co...
4. Customer Revenue in March: platform.stratascratch.com/co...
5. Monthly Percentage Difference: platform.stratascratch.com/co...
6. Premium vs Freemium: platform.stratascratch.com/co...
~~ Non-Coding ~~
1. Credit Card Activity: platform.stratascratch.com/te...
2. Outliers Detection: platform.stratascratch.com/te...
3. Probability of Having a Sister: platform.stratascratch.com/te...
4. Uber Black Rides: platform.stratascratch.com/te...
5. Terabyte of Data: platform.stratascratch.com/te...
The skills that we work on in this video include:
- Python Pandas
- Groupby & Aggregate DataFrames
- Use regexes to analyze text
- Datetime objects in Pandas
- Filtering by Conditionals
- Applying a lambda function to a data frame
If you have any questions, let me know in the comments!
If you enjoyed this video, make sure to throw it a like & subscribe for all future content :)
-------------------------
Video Timeline!
0:00 - Intro & Video Overview
0:46 - Check out this Video’s Sponsor, Brilliant!
3:10 - Coding #1 (Microsoft, Easy) - Finding Updated Records
10:36 - Coding #2 (Airbnb, Easy) - Number of Bathrooms and Bedrooms
16:38 - Coding #3 (Google, Medium) - Counting Instances in Text
28:23 - Coding #4 (Meta/Facebook, Medium) - Customer Revenue in March
36:51 - Coding #5 (Amazon, Hard) - Monthly Percentage Difference
56:38 - Coding #6 (Microsoft, Hard) - Premium vs Freemium
01:10:28 - Non-Coding #1 (Visa, Easy) - Credit Card Activity
01:13:33 - Non-Coding #2 (IBM, Easy) - Outliers Detection
01:16:46 - Non-Coding #3 (Google, Medium) - Probability of Having a Sister
01:27:19 - Non-Coding #4 (Uber, Medium) - Uber Black Rides
01:36:57 - Non-Coding #5 (Capital One, Hard) - Terabyte of Data
01:46:41 - Video Conclusion & Recap
-------------------------
Follow me on social media!
Instagram | / keithgalli
Twitter | / keithgalli
TikTok | / keithgalli
-------------------------
If you are curious to learn how I make my tutorials, check out this video: • How to Make a High Qua...
Practice your Python Pandas data science skills with problems on StrataScratch!
stratascratch.com/?via=keith
Join the Python Army to get access to perks!
KZbin - / @keithgalli
Patreon - / keithgalli
*I use affiliate links on the products that I recommend. I may earn a purchase commission or a referral bonus from the usage of these links.
This video was Sponsored by Brilliant

[@KeithGalli]

Thank you Brilliant for sponsoring this video! Check out brilliant.org/KeithGalli/ to get started learning STEM for free, and the first 200 people will get 20% off their annual premium subscription. Hope you all enjoyed this video :). I'm working on a bunch of new content right now so be on the lookout for another video or two in the next couple of weeks. If you have any questions about the topics covered in this or have a request for a future video, let me know here in the comments!!

[@edwardj.warden5072]

Hi @KeithGalli. I’ve got two questions to ask you. I have watched lots of your videos that I like, and learned a lot. My question is do you think that the certificate that Datacamp provides for data science is worth to earn, and would it help me to find a data science job? And, what best place, you recommend, in online to get certificate for data science that would help me to find a data science job? Thank you.

[@yogeshuttekar8542]

Glad to see you back mate. I have really learned more from your videos than attending University.

[@hardiktyagi1955]

At 37:48 I work for Amazon's RPA team, trying to make a career in data science. Last month I was appearing for an IJP and got the same question in SQL coding round. Thanks for making this Keith. Keep them coming.

[@KeithGalli]

Dang that's too funny. My hope is that this video will help people in similar situations to yours moving forward. Thanks for watching!

[@shivamburnwal7765]

Hey Hardik, can you tell me why exactly you are trying to make a career in Data Science? Is it because RPA doesn't have a good future in the industry or it is because you personally prefer the Data Science field. I am asking this question as I am also starting as a member of EXL's RPA team.

[@deepaksaikumar5178]

Hi Keith, You have been a great resource to learn Python and Data science-related skills. Thank you!

[@nicholasgrandizio7596]

Thank you for all the hard work you put into teaching Data Science. Your videos and others like you, provide more to the community such as myself trying to build a career in data than what University Programs provide. Your playing an important role in the future of Data Science by leading current students along the path to future industry leaders.

[@danielefarotti1061]

I really like your approach in explaining things. I am currently transitioning from pure maths into data science, and I find these videos very helpful!

[@BOGABOOfull]

Glad you're back bro ;) love this types of vids. Love from Portugal

[@Lnd2345]

Here's a one liner chained version I've come up with for coding #6 df = ms_user_dimension.merge(ms_acc_dimension, on = 'acc_id').merge(ms_download_facts ,on ='user_id').pivot_table(index = 'date',columns = 'paying_customer',values = 'downloads',aggfunc ='sum').reset_index().query('no > yes')

[@masked00000]

You're literally the best tutor I have seen, I myself am a Data Scientist but the amount of data science approaches I learn from you is incredible, I started from your channel and always wait for you to post new video, Hat's off. Love from Pakistan.

[@laurentreynaud4404]

Thank you so much for these data science courses!

[@netanelmad]

Thanks for the video! Would love to see your approach to more non-coding questions specifically :)

[@niteshprajapat7918]

You are gem ❤️ the way you explain concepts are at next level 🔥🔥

[@dinkinflicka157]

Yay! Another real world problem solving video. Thanks Keith. Love your content as always.

[@KeithGalli]

Glad to hear it, I appreciate your support!! :)

[@xxxihabxxx1]

this took me a week to finish all coding questions, 10000% helped me alot to practice everything i learned in your previous pandas crash crourse. thanks

[@troy671]

Thanks for the video. It is great to see your thinking process even though you are not an expert in pandas.

[@edwardj.warden5072]

Very helpful. Thank you Keith.

[@adeafni9544]

Thank you Keith, you're amazingg, keep it up!!!

[@expat2010]

I really enjoy the real world feel of your videos. Probably now ChatGPT would be a lot faster than searching Stackoverflow or the Pandas docs for those things that one doesn't know by heart.

[@kennethstephani692]

Great video, Keith!

[@arashomranpour5468]

good having you back

[@KeithGalli]

Good to be back! :)

[@ansekao4516]

Great video, please do more like that. Watching you for a long time

[@wiz8058]

Great work man!! you're always doing the best.🔥🔥🔥

[@KeithGalli]

Thank you for the support as always!!

[@zanerios2776]

really love the style and format of vid, just subbed

[@KeithGalli]

Glad you liked it man! Thanks for the sub

[@kumaripritika2799]

Really helpful video!

[@iamTHIEN013]

Hi Keith , Thank you so much for these videos, could you make more videos about power PI or Tableau, really really appreciate it .

[@a.5214]

amazing! we want more of this stuff 👌

[@KeithGalli]

Appreciate it! More coming soon :)

[@mekuzeeyo]

Thank you for coming back🤗

[@KeithGalli]

Happy to be back!!

[@phoenixcollege6608]

makes it easy to understand watching your vid on a friday night and these are the best years of my young life

[@bobbyg603]

Glad you're back bro!

[@KeithGalli]

thanks brother!!

[@wahaha108]

long time no see keith, welcome back 😀😀

[@user-xj9re7gv5g]

It is very great. Thank You!

[@udayabhaskar1495]

Thank you for this video!👍

[@n_12346]

Brilliant video! very helpfil

[@9eartheyes]

great video! thank you!

[@pratikpawar336]

great video, please make more video like this

[@iamfavoured9142]

Welcome back Keith 💃🏻💃🏻

[@prof_albert]

That was great. Bravo and all of your videos are awesome 🌺👌💞🤩💪

[@ranjithraghunathan1267]

Thanks Keith

[@DendrocnideMoroides]

yes please make more videos like this

[@mehdismaeili3743]

excellent, thanks.

[@KeithGalli]

You're welcome :)

[@phsopher]

For the fifth problem, pandas has an in-built percentage difference method (pct_change). The solution could be as follows for example: sf_transactions['year_and_month'] = sf_transactions.created_at.dt.strftime("%Y-%m") monthly_revenue = sf_transactions.groupby(["year_and_month"]).sum().reset_index() monthly_revenue['pct_change'] =(monthly_revenue.value.pct_change()*100).round(2) monthly_revenue[['year_and_month','pct_change']]

[@KeithGalli]

Oh cool, I didn't know that! Thanks for sharing :). Nice solution 🤠.

[@RahmanIITDelhi]

Hey ,Keith ..Can we access library during the solving at real time exam?

[@jovanjanjic9029]

In question #3 Counting Instances in Text you should add filters=re.I to account for capital letters: len(re.findall(r'\bbull\b', text, flags=re.I)))

[@jovanjanjic9029]

Great video btw!

[@fcoatis]

Great video Keith. I just got curious how you comment a block of code?

[@kinghezzy]

Highlight and ctrl+/

[@drakkarleon]

Yeeeeeeeeyyy!!!! i love your enthusiastic cry of success :D 26:31

[@user-if1dj7fy2y]

Bravo 👏 Lit 🌠 Impressive 👌 ❤ Gratitude 🥳 for your satisfactory Work 💪🚀💯💪

[@fantasyxpress7966]

Is dsa important for data scientists too keith

[@anonviewerciv]

That first one and others are SQL problems converted to pandas. I suppose that's a decent way to get basic pd questions. (28:48) 17:20 I know it's more a reference to the stock market terms, but I can't stop thinking of Fallout: New Vegas. 1:11:00 If you have the locations that's just a simple matter of putting it on a map and seeing where it clusters the most. 1:28:00 Context, context, context. Was that the only reduction?

[@konstantinpluzhnikov4862]

These stratascratch tasks could be solved in sql. The site provides this option.

[@AIdevel]

The problem lays in your use of round function you supposed to wrap the equation with round and then select the decimals 2

[@user-zm6kj7oi3d]

you are helping a high schooler out by being back

[@KeithGalli]

More videos coming soon :)

[@finnnelson5472]

TY :)

[@MikeResurrected]

Could you actually google for help during a DS coding interview nowadays?

[@ranjithraghunathan1267]

how can i download or copy the raw dataset for each part ?

[@vanshmalik1446]

Hey! Does anyone knows more of the data analysis pay after placement programs accepting applications all over the globe?

[@AIdevel]

Replace yes with 1 and no with zero and sum them

[@konstantinpluzhnikov4862]

LifeHack: if you are short of money, but want to use a service, use vpn of relatively poor country. Result will be interesting.

[@DendrocnideMoroides]

did you ever use it? and on which website?

[@balakumar.n4891]

super

[@manphu2515]

Thanks so much for the video, learn a lot from you. And you are super cute 😍

[@YunusFidan_]

Noice!

[@meujie8835]

Hi, I'm Jiemeu and I love your channel. I hope to discuss business cooperation with you.....

[@jovanjanjic9029]

Your solution for the Probability of Having a Sister question is not correct. We know for sure that the random girl must be from the [1, 2, 3, 4] part of the dataset, which amounts to 0.7. We should divide the probabilities for 1, 2, 3, 4 with 0.7, to get the probabilities that the girl is from each of these families. She theoretically can't be from families with 0 and 5 children. Essentially, you are counting in the possibilities of she being in families 0 and 5, even tough it's impossible. (In practical terms, you are needlessly being blind about the info you already have.) So the correct solution is: 0.25/0.7 x 0 + 0.2/0.7 x 0.5 + 0.15/0.7 x 0.75 + 0.1/0.7 x 0.875 = 0.42857, which is 0.43 when we round it up.

[@doulaishamrashikhasan8425]

you disappeared again 😢

[@KeithGalli]

My apologies! I have a video that I'm finalizing the editing for. It should be out in the next 3-4 days and then I'm going to try to be more consistent!!

[@ratkillerthe]

I solved the Bathrooms/Bedrooms problem with: cols_of_interest = airbnb_search_details[['city', 'property_type', 'bathrooms', 'bedrooms']] property_results = cols_of_interest.groupby(['city','property_type']).agg( avg_bathrooms = ('bathrooms', 'mean'), avg_bedrooms = ('bedrooms', 'mean')).reset_index()