ML Interview questions

Given an m x n matrix board containing 'X' and 'O', capture all regions that are 4-directionally surrounded by 'X'. [Had to check solution, try again later]

How would you show the user "the best value for money" hotels? How would you evaluate your solution? what features would you select to explain the "value"? What are the assumption of a linear regression model? What would you do if the relation between outcome and features is not linear? How do you validate the model you built? Design and describe an experiment to confirm that the method you developed is a good one.

https://booking.ai/

Complete modeling part of FB engagement quesition

Read: https://booking.ai/https-booking-ai-machine-learning-production-3ee8fe943c70
My summary:
https://psvishnu.substack.com/p/summary-bookingcoms-rs-the-machine

🛐 One of the Core ideas of booking.com: "Diversity gives us strength".

RS, the Machine Learning Productionization System

🥕Requirements
1. Consistency
2. High availability
3. Low latency
4. Scalability
5. Observability
6. Reusability

👨‍👨‍👦‍👦 The fantastic four approach
▴ Lookup tables:
▴ Generalized Linear Models (GLMs)
▴ Native libraries
▴ Scripted models

Kaggle

▴ Revisit: Opencv lane detection
▴ Complete: Cassava classification article
▴ Revise: CNN, Object detection work

🤔 How would you improve engagement on Facebook?

💭 Spam mail classifier: Enron email dataset
💭 Tone detection similar to grammarly
💭 Sentiment analysis

- CASE statement
- Distinct
If you include two (or more) columns in a SELECT DISTINCT clause, your results will contain all of the unique pairs of those two columns.
- Union and Union ALL, Union will remove all the duplicates whereas UNIONALL will keep dups.
- CAST(column_name AS integer)
- Date functions:
companies.founded_at_clean::timestamp +
INTERVAL '1 week
- Window functions like PARTITION, LEAD, LAG, NTILE

Next: String functions & Date functions, Indexing, Remaining window function

- Need to take AWS/GCP/Azure & REST APIs test on Linkedin
- Read more about distributed training
- CI/CD pipelines for E2E ML Use-cases

🏋🏻‍♀️ SQL Revision (Basics -> Intermediate)
• Comparison operation even works for letters. eg: where month_name > 'J' (This will include January). But NULL operations are not valid like greater than check for the null column.
• ILIKE is a logical operator for case insensitive matches.
• IS NULL check.
• COUNT(column_name) will give NOT NULL and NON UNIQUE values.
• COALESCE( .... ) any number of arg returns the first NOT NULL argument.
• WHERE doesn't work with aggregate columns that's where the HAVING clause comes in
• If-then equivalent is CASE- WHEN-THEN-ELSE[Optional]- END in SQL. ⭐️ It can be used inside SELECT, GROUP BY, Inside aggregate functions,
COUNT(CASE WHEN year = 'FR' THEN 1 ELSE NULL END)
• Declare Variables in SQL using SET @Variable1 = 'test'

🥊 Questions
• Write a query that returns all rows for songs that were on the charts in 2013 and do not contain the letter "a".
Ans: NOT ILIKE '%a%'
• How to insert a slice of a table into another table?
Ans: Insert into table (col1, col2 ...) values (select col1, col2... )
• Write a query that counts the number of 300lb+ players for each of the following regions: West Coast (CA, OR, WA), Texas, and Other (everywhere else).
Ans: group by case ... end

Next: https://mode.com/sql-tutorial/sql-case/

git: DS questions/ML

Every time I prepare for Interviews this is what I feel