How To Answer Complex Questions IN SECONDS! (Fermi Estimates)

Data analytics and data science interviews for companies like Google, McKinsey, Amazon and more always ask Fermi Problems. But why? How can you learn how to solve them and get ahead in your data career. In this tutorial find out how Fermi estimates combined with a knowledge of landmark numbers can help you get further in your career.
00:00 - What are Fermi Problems?
00:42 - Why it's important to be able to answer them
01:11 - Why Fermi estimates works so well
01:34 - Example 1 and answer
02:19 - Example 2 and answer
03:36 - Example 3 and answer
04:55 - How to use Landmark Numbers

[@iprobablyhaveanameofsomeki7698]

Thank you so much for those landmark numbers, they’re incredibly helpful for Fermi questions!

[@undatable5838]

Glad you like them!

[@sebastianschmidtataktien-e4705]

I like the more complex ones. If there are fewer interim estimates required, then you may be further off with the estimate as you will have less benefits from overestimations and underestimations evening out. Fun stuff. I estimated way fewer with the blades of grass. 36.000.000. You will do best if you have several people doing a Fermi estimate and then averaging.

[@CreateurWealth]

you are amazing- Hats off your all your effort and creativity... i loved all your videos

[@undatable5838]

Thank you, that’s very kind of you! ☺️

[@beautifulsmall]

UMIST physics question, for clean drinking water at the equator, de-salinate or drag icebergs. My answer was about the same. Kg m / s is all you need. Combine this monte carlo with fast aI. Scale is everything, Giga Farad to atto second

[@physicsbutawesome]

Nice, I have always found Fermi questions fascinating. Have you ever been asked one in real life, like an interview?

[@undatable5838]

I have indeed. Earlier in my career. And I completely messed it up. I didn’t know what Fermi questions were back then, so I was more concerned that the rough numbers I was using were wrong, and lost all confidence! 🤣 That’s why I like Fermi’s principle about under/overestimates potentially cancelling each other out, and the idea of having come ‘landmark numbers’ that you can always refer to. PLUS I now know that, in interviews, the people interviewing you are more interested in the logic you use to get to the answer than the actual answer itself 😉

[@collinguckin5225]

20 blades of grass in a cm² is way over. I might say around 5-10

[@undatable5838]

Yeah, I thought afterwards 20 was too much. Not sure I’d go down to 5 though… what would your end estimate be?

[@thomasadams5078]

@@undatable5838 When I tried this problem, before listening to the solution, I drew the 1cm x 1cm square and estimated that 8 blades of grass would fit in it by drawing it :)

[@idiot528]

@@thomasadams5078 in fermi estimations you should go with very rough estimates to for easy solutions

[@collinguckin5225]

@@undatable5838 idk I'm gonna go count real quick. Glad we're finally getting to the bottom of this

[@bot.]

@@collinguckin5225 Done counting?

[@tharun541]

Great video. Any resources for developing skills in Fermi problems?

[@undatable5838]

Hi Tharun, the best thing you can do is to practice as much as you can, until it becomes a habit. If you Google ‘Fermi problems’, there are some fantastic educational resources that schools have developed for their students… and regardless of your age, they’re worth giving a go.

[@marvelz6355]

Awesome vid thanks. That poor couch though haha

[@undatable5838]

🤣 Yeah, I think I may have broken it! 🤣 I now need to find a new way of injecting energy into my voice! Glad you liked it, and thanks for the comment.

[@near9831]

Thank you very much i really like this video of you ( sorry my english is bad)

[@undatable5838]

I'm glad you like it!

[@arjunrajkumar6826]

Intriguing! Is this actively being used in any field? Or is it just an abstract idea?

[@undatable5838]

Hi Arjun, yes it’s used quite a lot. It was named after one of the world’s most famous physicists, so that’s a clue. But also it’s a shortcut used by a lot businesses, engineers and others to either give an estimate, or help to inform what additional data/exploration is needed. It’s not recommended if you have a life or death situation… but if you’re time limited or lacking data, then it’s a fantastic technique! Hope that helps! ☺️

[@stephanddd]

All the time. Frequently called a "guesstimate" :) And even if it's not used to actually derive the actual answer, this approach can be used to confirm that the calculated answer makes sense. From buying groceries to producing a business plan, to calculating the weight of the materials for a bridge, to figuring out how many developers you need in a startup.

[@MaxVenous]

I came to know about this Fermi thing From an Anime 🌚 - Amagi Brilliant park , it's something about a guy trying to re-establish a broken amusement park

[@undatable5838]

Love that Max! Had no idea. Googling it now… ☺️

[@ukaszsek1684]

1st exp. i guestimate 10 blades of grass for each 1cm3 i guestimate field for 30 meters by 70m - numbers of blades of gras approx 200 milion ( correction after checking: field is 100x70m so i underestimate somwere 3,5 times :P but still close :))

[@undatable5838]

🤣 so glad you did this and shared it! Thanks!

[@planktonfun1]

and your teacher always says wrong because you didn't use it her way

[@undatable5838]

🤣 at least you know the truth!