4 ms·

Data Scientist simply means a developer who also knows statistics. Here's how to sniff them out: 1. What is you favorite programming language and why? 2. Whic

by snorkel 14y ago

Data Scientist simply means a developer who also knows statistics. Here's how to sniff them out: 1. What is you favorite programming language and why? 2. Which is your least favorite programming language and why? 3. Explain how Bayesian spam filtering works. 4. How do you determine if a given data sample is statistically significant? 5. Suppose we ran a brand advertising campaign on radio and on television. Neither ad campaign uses special tracking codes or custom landing pages, both ads simply mention the web site address. What tools and methodology would you use to measure the response rate from radio ads vs. tv ads, and predict the total response that will generated from each ad?

a_bonobo 14y ago
>4. How do you determine if a given data sample is statistically significant? That's a weird question - what am I to prove? Are there differences in the dataset? How does the data look like? Is there a second dataset against which to compare? What kind of data is it, ordinal or nominal? Is the dataset normally distributed? I can't just say "that data sample is statistically significant" without a background against which to compare against! >3. Explain how Bayesian spam filtering works. I would say the majority of non-web developer scientists don't know how this works, why should we?
- tymekpavel 14y ago
  If you've had any exposure to Bayesian statistics, it's very easy to explain the approach. You're basically tokenizing the email by word, and then predicting the probability that P(email is spam|token). I would expect any data science candidate to have (at least) a basic understanding of Bayesian statistics.
  - a_bonobo 14y ago
    It's just that an e-mail filter is the last thing most scientists working in biology, chemistry, physics, medicine are actually working with, so it's (in my opinion) a rather unfitting example, you might be selecting against these people who fit the bill of "data scientist" probably better than the average "I once did something in Excel"-web guy.
dchichkov 14y ago
These questions would get you: recent grads, who have not forgotten all the math; hackernews crowd, who reviewed pg research from years back; developers, who have taken some popular AI/ML class. And while this is all good, this is certainly not enough. In reality applied data science is not different from any other area. You get proficiency with experience, as usual.
arrrg 14y ago
Question 4 makes no sense.

by snorkel 14y ago

a_bonobo 14y ago

>4. How do you determine if a given data sample is statistically significant? That's a weird question - what am I to prove? Are there differences in the dataset? How does the data look like? Is there a second dataset against which to compare? What kind of data is it, ordinal or nominal? Is the dataset normally distributed? I can't just say "that data sample is statistically significant" without a background against which to compare against! >3. Explain how Bayesian spam filtering works. I would say the majority of non-web developer scientists don't know how this works, why should we?

tymekpavel 14y ago
If you've had any exposure to Bayesian statistics, it's very easy to explain the approach. You're basically tokenizing the email by word, and then predicting the probability that P(email is spam|token). I would expect any data science candidate to have (at least) a basic understanding of Bayesian statistics.
- a_bonobo 14y ago
  It's just that an e-mail filter is the last thing most scientists working in biology, chemistry, physics, medicine are actually working with, so it's (in my opinion) a rather unfitting example, you might be selecting against these people who fit the bill of "data scientist" probably better than the average "I once did something in Excel"-web guy.

dchichkov 14y ago

These questions would get you: recent grads, who have not forgotten all the math; hackernews crowd, who reviewed pg research from years back; developers, who have taken some popular AI/ML class. And while this is all good, this is certainly not enough. In reality applied data science is not different from any other area. You get proficiency with experience, as usual.

arrrg 14y ago

Question 4 makes no sense.