Restricted category To view older posts, you must first submit an approved post in that category

Linked In Data Science Interview Experience

First of all - the process with LinkedIn was sloooooooow :crazy_face: They were my top company and I’m glad I applied there first because it took forever. I applies in the beginning of August and interviewed in September. It took a few weeks to tell me how I did also, so they had me all nervous!

3 parts of my interview

  1. Case study: what to do if the application rate drops by 10%
  2. Stats & ML: linear regression assumption, regularization
  3. SQL: Two tables, one is the status of the current month, the other is the activities of the business, goal is to the status of the current user

TABLE 1: “status”-contains all LI members’ latest push notification setting status as of the last day of Jan (01/31/2020)

member_id status

  1 on
  2 off
  3 on
  4 off

TABLE 2: “actions”-all actions that members made in Feb (after the time period of ‘status’ table).
(For simplicity, They let me assume a member can have at most one action per day)
member_id date_sk action

  1 2/2 turn_off
  1 2/5 turn_on
  2 2/3 turn_on
  4 2/10 turn_on
  4 2/13 turn_off
  5 2/13 turn_off

EXPECTED RESULT-the current status (as of 02/29/2020).
member_id current_status

  1 on
  2 on
  3 on
  4 off

Summary of the onsite:

  • Hiring manager : Behavioral & projects in resume
  • Data imputation: SQL & R/Python
  • Case study: about LinkedIn products
  • Stats & ML, AB testing, hypothesis testing, probability
  • Storytelling: analyze the data and make a presentation
3 Likes

Can you share about the case study?

Mine was " What would you do is LI signups drop by 10%?"
Sorry if not clear in the post

That sounds like some of the Product Execution questions I got for Facebook - curious how the answer differs in data science???

2 Likes

Following this post—just now applying to LI for DS.

1 Like

Much more data/numbers driven i beleive. I’m not a PM so I don’t know what is expected but DS is definitely quantitative.