I did my first interview and the next step is an assessment. I received a "big data set" and the task only says "We would like to know what clusters there are using the provided data set". I asked them to send the test on a Friday as I only have time during the weekend but they sent it already Monday. The messaged ended "Please let us know when you are ready to present".

Cluster analysis is quite the never-ending task and there is no "correct solution" and I've spent about 10h this weekend, but I still have lots of ideas of things to test, which will require analysis and additional time. When should I be satisfied and should I include everything I did?

Tomorrow I want to propose that I present on either Friday or Monday because that is the only day I'm available to present. It does however mean that I will have had the test for almost 2 weeks. Will the hiring team increase the "threshold" because I've had the test available for such a long time?

  • 1
    This is not possible for us to answer. Commented Jun 29, 2020 at 4:18
  • I think we can view this question has "i received a technical assessment for interview that take quite some time and there is no 'ONE solution'" and I don't have a clear view of what they expect as a solution, how should I deal with it ? This is at least the angle I choosed in my answer.
    – Walfrat
    Commented Jun 29, 2020 at 11:01
  • @JoeStrazzere No, would have been a smart thing to do.
    – MLEN
    Commented Jun 29, 2020 at 18:17
  • As a professional data scientist, when would you be satisfied with your answer? No doubt, they are looking to if your expectations are aligned with theirs to see if you are a fit for the job.
    – Seth R
    Commented Jun 29, 2020 at 19:38

3 Answers 3


Cluster analysis is quite the never-ending task and there is no "correct solution" and I've spent about 10h this weekend

Most of data science topics are like that, aren't they? The models can always be improved and are never perfect. So don't mention that when contacting them.

Actually I hear alarm bells when someone tells me something is a never-ending task, since I've had team members with this approach. They all wanted to spend weeks working on "never-ending tasks", just experimenting and not being interrupted in the process. And only communicating after weeks of solitary experimenting around.

This can work in academia, but not in business. Business is about deliverables. In business you need to be able to say that e.g. you see 5 main approaches to the task and you first want to focus on approach A since whatever. Then you would go for B because whatever. C, D and E would, in your opinion, be the last ones to consider.

Just give them an overview of how you would approach the topic and a practical example of what you would do.

What you received is quite a typical example of doing DS for a non-research institution. The CMO comes to you and says: "I want to have my customers segmented. How will you do that?", so if you don't like working like that, you should think whether applying for such jobs makes sense


It is impossible for anyone other than the potential employer to guess. You realistically need to clarify any requirements with the company. Ideally, you would have replied on Monday when they sent the test asking something like

Cluster analysis is obviously something that can take quite a bit of time and doesn't have an obvious endpoint. How much time do you expect a qualified candidate to spend analyzing the data before the presentation?

Spending 10 hours, assuming you're qualified for the position, seems like a lot for an assessment barring exceptional circumstances (i.e. you know that they've already narrowed it down to a handful of candidates, the position is exceptionally attractive, etc.). It is pretty much always the case that you can spend a nearly infinite amount of time polishing your approach. Ideally, the company is just expecting a few hours worth of work to get an idea of how you approach the problem but plenty of companies expect what I'd consider an unreasonable amount of effort in these assessments.


Honestly, it could be different cases :

  • Maybe they're trying to get some work out of you for free.
  • They're just interested in raw results.
  • They're more interested in your methods to find the results and also also how you define that your solution is good enough.

Excluding first option : if you need a job for money you may answer considering they want raw results. Now if your searching a company with a decent organization that know how project based on classification algorithm/AI works, I would answered based on my last point. If their answer like "95% good result is not enough we want 100%" you may think twice before working here.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .