自动车Waymo数据科学DS/DA OA

【求米】LZ大米都烂光了 T_T题都看不到，盲冲的OA，求帮转好人一生平安

• 4道SQL
• 2道Python
• 2道Free response
• 整体时间1.5h，很紧凑
• CodePad

您好！
本帖隐藏的内容需要积分高于 188 才可浏览
您当前积分为 0。
使用VIP即刻解锁阅读权限或查看其他获取积分的方式

游客，您好！
本帖隐藏的内容需要积分高于 188 才可浏览
您当前积分为 0。
VIP即刻解锁阅读权限 或 查看其他获取积分的方式
Unlock interview details and practice with AI
Curated Interview Questions from Top Companies

被莫名加了隐藏，这是隐藏内容
Q1 - SQL: Inactive User Analysis
  Count users with zero rides in the 7-day window ending July 7, 2024.
  Tables: rides(ride_id, ride_date, ride_rating, user_id), users(user_id, city).
  Must handle users who signed up but never took a ride.

  Q2 - SQL: Monthly Ride Aggregation
  Compute total rides and average rating per month, rounded to 2 decimals, sorted by month descending.
  Tables: rides(ride_id, ride_date, ride_rating, user_id), users(user_id, city).

  Q3 - SQL: Low Frequency Users
  Count unique users who have taken 0 or 1 total rides, grouped by city.
  Tables: rides(ride_id, ride_date, ride_rating, user_id), users(user_id, city).
  Must include users who never took a ride.

  Q4 - SQL: Retention Rating Analysis
  Compute average ride rating for the 1st and 3rd "active month" of usage, grouped by city.
  Active month = Nth distinct calendar month a user had rides (skipping inactive months).
  Average is across all rides (not average-of-averages).
  Tables: rides(ride_id, ride_date, ride_rating, user_id), users(user_id, city).

  Q5 - Python: Average Function
  Write average(table) that returns the mean of a list of numbers, or 0 if empty.

  Q6 - Case Study: "Smart Wait" Launch
  Waymo launched "Smart Wait" to improve ETA accuracy, showing more conservative (longer) wait times.
  Results: conversion rate dropped 5%, actual TTP decreased 25%.
  (1) A PM says fleet efficiency improved. Do you agree? Why or why not?
  (2) Alternative hypotheses for the TTP decrease and how they relate to the conversion drop?

  Q7 - Python: A/B Test P-Value
  A/B test in SF, July 2024 for a new routing algorithm to reduce TTP.
  Given df_users(user_id, variant) and df_rides(ride_id, user_id, ride_date, city, time_to_pickup).
  Filter to SF + July 2024, join with variants, run Welch's t-test on TTP, return p-value.

  Q8 - Statistics: Experiment Validity (follow-up to Q7)
  (1) Based on the p-value from Q7, are results statistically significant?
  (2) Is this conclusive proof the new algorithm is better? If not, describe a better experiment design.