Skip to contentSkip to footer

Data Scientist Interview Questions

Salaries
Interviews

Data Scientist Interview Questions

In un colloquio per Data scientist, ti verranno poste domande volte a verificare le tue capacità di data modeling, risoluzione di problemi e programmazione. Preparati a rispondere a domande di carattere generale che valutano la tua conoscenza della statistica e della scienza dei dati. Dovresti inoltre prepararti a rispondere a domande aperte mirate a testare la tua creatività, le tue doti comunicative e la tua formazione nella programmazione e modellazione dei dati.

Domande tipiche dei colloqui per Data scientist e come rispondere

Question 1

Domanda 1: Quali tecniche di data modeling preferisci e perché?

How to answer
Come rispondere: Trasformare i dati in informazioni comprensibili e fruibili è un aspetto fondamentale del lavoro di Data scientist. Con questa domanda i datori di lavoro vogliono capire il tuo backgruond e valutare le tue capacità di data modeling. Elenca e illustra le tecniche di data modeling che preferisci, includendo vantaggi come semplicità d'uso, flessibilità, ecc.
Question 2

Domanda 2: Come rilevi gli account Instagram fasulli utilizzati per raggirare i consumatori?

How to answer
Come rispondere: Domande come questa permettono ai selezionatori di testare le tue capacità di risolvere i problemi. Quando rispondi a domande aperte di questo tipo, non esitare a chiedere chiarimenti o a usare lavagne per dimostrare le tue abilità nel tracciare diagrammi e usare codici. Condividi il tuo processo di pensiero mentre elabori il problema.
Question 3

Domanda 3: Descrivi quali circostanze richiedono una lista, una tupla o un set in Python.

How to answer
Come rispondere: I selezionatori ti porranno domande come questa per testare le tue abilità di programmazione in Python. Ripassa gli elementi fondamentali di Python, come liste, tuple e set prima del colloquio. Dovrai essere in grado di spiegare quando e come ogni strumento deve essere usato da un Data scientist.

54,319 data scientist interview questions shared by candidates

1. Started with a detailed explanation of a past project - what was the business question, how did you come up with the solution, what was your hypothesis, how did you design the A/B test, why did you make certain choices, what was the result etc. Prepare 1-2 examples from your past, where you can talk in depth about the technical elements of your project. 2. Let's say we have a dataset with attributes for a house (Sq footage, locality etc) and house price. How will you predict the house price from these attributes? (Build a multiple regression model) 3. For this multiple regression model, explain the end-to-end process. What steps will you take before building the model, how will you impute missing values, how will you handle outliers etc. What are the underlying assumptions of a regression model? 4. Once the model is built, how will you infer the relationship (sign and magnitude) between the house attributes and house price. How will you explain it to someone that's not a technical person? 5. For the regression coefficients, how will you interpret them, (p-values, confidence interval etc). How will you explain a p-value to a layman 6. Next question was about "how will you segment customers" in order to serve a business requirement, such as determining which customers to show a given ad (I answered with clustering, because the business problem wasn't very specific, he just described it very generally) 7. For clustering, how does it work, how to choose the value of K in k-means. I also said we can use Gaussian mixture models for clustering, which he didn't seem to know because he asked me to clarify what I mentioned. There might have been a few more questions that I don't remember, but the theme of the interview was to check how well you know the basics of Stats/ML. I believe I answered most of the questions correctly so to receive the feedback that I wasn't up to the mark technically seemed like a case of Google not wanting to reveal the real reason, whatever it was. Either way, make sure you confirm the format of the interview with the recruiter. Because I was already interviewing with other companies, I had brushed up on my Stats/ML basics, but you might not be as lucky. Good luck!
avatar

Marketing Data Scientist

Interviewed at Google

4.4★
Nov 19, 2020

1. Started with a detailed explanation of a past project - what was the business question, how did you come up with the solution, what was your hypothesis, how did you design the A/B test, why did you make certain choices, what was the result etc. Prepare 1-2 examples from your past, where you can talk in depth about the technical elements of your project. 2. Let's say we have a dataset with attributes for a house (Sq footage, locality etc) and house price. How will you predict the house price from these attributes? (Build a multiple regression model) 3. For this multiple regression model, explain the end-to-end process. What steps will you take before building the model, how will you impute missing values, how will you handle outliers etc. What are the underlying assumptions of a regression model? 4. Once the model is built, how will you infer the relationship (sign and magnitude) between the house attributes and house price. How will you explain it to someone that's not a technical person? 5. For the regression coefficients, how will you interpret them, (p-values, confidence interval etc). How will you explain a p-value to a layman 6. Next question was about "how will you segment customers" in order to serve a business requirement, such as determining which customers to show a given ad (I answered with clustering, because the business problem wasn't very specific, he just described it very generally) 7. For clustering, how does it work, how to choose the value of K in k-means. I also said we can use Gaussian mixture models for clustering, which he didn't seem to know because he asked me to clarify what I mentioned. There might have been a few more questions that I don't remember, but the theme of the interview was to check how well you know the basics of Stats/ML. I believe I answered most of the questions correctly so to receive the feedback that I wasn't up to the mark technically seemed like a case of Google not wanting to reveal the real reason, whatever it was. Either way, make sure you confirm the format of the interview with the recruiter. Because I was already interviewing with other companies, I had brushed up on my Stats/ML basics, but you might not be as lucky. Good luck!

Viewing 1541 - 1550 interview questions

Glassdoor has 54,319 interview questions and reports from Data scientist interviews. Prepare for your interview. Get hired. Love your job.