Allstate Corporation is an American insurance company that specializes in car insurance while also having personal lines of insurance. The Allstate data science program is well refined. Allstate is one of the few companies that has jobs for junior data scientists roles for non-experienced candidates or candidates coming out of school.
Allstate Data Science Interview Process
The first step depending on the level of the Allstate Insurance data scientist position will be to apply online. Allstate will send you an email if you pass the initial screen and ask you to call a number with ten automated questions while they record your response to those questions. These questions are mostly behavioral but it can be unnerving to be speaking to an answering machine so it helps to practice before-hand. A few of the general questions are about projects related to data science, salary expectations, programming languages you know etc..
Next up was a technical interview that took an hour. The technical screen consists of two problems: a problem solving case question and a probability and statistics question.
The most common probability and statistics question will be around pulling jelly beans from a jar. Given an unknown number of jelly beans in a jar, if you select a jelly bean one at a time, how many rounds will it be until you pick two of the same color. What about when selecting N of the same color jelly beans?
You are about to get on a plane to Seattle. You want to know if you should bring an umbrella. You call 3 random friends of yours who live there and ask each independently if it's raining. Each of your friends has a 2/3 chance of telling you the truth and a 1/3 chance of messing with you by lying. All 3 friends tell you that "Yes" it is raining.
What is the probability that it's actually raining in Seattle?
The case question will likely be about building a model for a transportation company. Allstate wants you to get to the bottom of what the major business questions are and describe a statistical model that would answer it. The case study presents real data and they ask questions about how you would approach the modeling process. Understand imbalanced class classification, model evaluation, cross validation, and logistic regression, and more advanced machine learning techniques.
The Onsite Interview
The Allstate data scientist onsite interview will be structured with an interview panel with 4 different interviewers. The onsite will last around 4.5 hours long with a one hour lunch in-between.
The data science onsite structure is broken down with:
- Behavioral questions
- Whiteboard coding and programming
- One hour lunch with two different employees
- Business and product case study
There may be a thirty minute job applicant led presentation as well depending on seniority of the position. Expect lots of questions prying at how deep your mathematical understanding of machine learning models are. Also be prepared for coding in Python and SQL and understanding how to solve basic algorithms along with complex joins.
The Allstate data scientist interviewers mostly want you to communicate your thought process and understand that you should think of the overall method before jumping into details. On the business case study, remember to list fundamental questions about verbalizing the business objective, selecting variables, transforming them, what to do if you’re missing data, and interpreting time series plots and saying what they’d mean for how the variables behave and if there are red flags with the variables.
Sample Allstate Data Scientist Interview Questions
- Given a dataset, how would you handle NULL values?
- Write a function to calculate the average given a set of integers.
- How would you calculate the number of draws you need to take so that you get 2 jelly beans of the same kind from a jar full of 3 kinds of jelly beans?
- Given a list of numbers: 10,20,30,40. What is the code to produce the output: 100, 90, 70, 40. What would be its time complexity?
- Given a list of floats, write a function to round exact values and not round inexact values.
- Write code to generate the K-Nearest Neighbor algorithm.
- Given two tables, one of customers and another of car accidents, write a SQL query to get the top car accidents by day.