I have a dataset which contains information about when do people enter and leave a premise. I have the following information in the dataset :

  1. Person Id
  2. Time of Entry
  3. Time of Leaving

The dataset has around 50 unique persons. Each person will have multiple entries corresponding to multiple visits. The data spans over a year so I have quite a lot of entries (around 1 million).

These people can be classified on the basis of the department they work under (2 departments - mutually exclusive) or on basis of their role (4 possible roles - all mutually exclusive)

I was wondering what kind of data analysis can be done with this kind of dataset. I am not looking for straight-forward things like "who spent the most time in building". However things like finding correlation between visits of 2 people would be interesting. So if person A visits the premise, what is the probability that person B would also visit. Since I have only around 50 unique visitors, I think such an analysis is feasible.

Another line of thought was to apply some interval-pattern mining techniques but I am not much familiar with them.

Can someone give me some pointers/ideas about what kind of data products can be build using this or what kind of techniques can be used with such data.

Edit : As discussed in comments, I call it a product in the sense that I do not want some simple or trivial analysis. And I am not looking for any commercially viable idea - just some cool fun idea :)

  • $\begingroup$ When I read the question's title, I thought you expected some commercially viable product ideas -- something probably outside the scope of this site. But you seem to be looking for some interesting analysis. Why do you call it a product? What is it you expect as an outcome? $\endgroup$ – logc May 28 '15 at 11:04
  • $\begingroup$ I call it a product in the sense that I do not want some simple or trivial analysis. And as you correctly mentioned - I am not looking for any commercially viable idea - just some cool fun idea :) $\endgroup$ – Shagun Sodhani May 28 '15 at 12:12
  • $\begingroup$ @Shagun This is an open ended question that is not suited for this format. What seems trivial to you, might be involved for someone else and high return generating (for a product, app or an organization). The answers to this question only encourage meandering discussions not suitable here (it also seems unlikely that you will ever accept an answer as its going to be hard to judge if this question has been appropriately answered). Please consider moving this question to another forum. $\endgroup$ – Nitesh Jun 1 '15 at 21:19
  • $\begingroup$ @Nitesh You are right about accepting any of the answers as it will be difficult to judge. But as I stated in the question itself, I am not looking for a revenue generating product. I am just looking for what possible stuff can be done with a typical dataset like this. You are correct when you say this is open-ended but then I am not expecting people to explain all of their approach. If the answer mentions a relevant keyword or a pointer towards the answer, I would be upvoting it and would consider it a valid answer. I would move it to another forum if you feel the points are not valid :) $\endgroup$ – Shagun Sodhani Jun 2 '15 at 4:10
  • $\begingroup$ Unfortunately, this question-answer format does not encourage idea generating discussions (for example, the possible things that can be done with this dataset) even though I personally like to indulge in such discussions :( $\endgroup$ – Nitesh Jun 2 '15 at 18:42

You might be able to find individuals that act as anchors. That if Abe is here, then everybody else is more likely to stay longer. Interesting to know if it was always the same people who stayed or not.

There might also be some interesting patterns with groups that arrive, are present, or leave together.

As a business owner you could look for individuals who not just work shorter hours, but arrive at their own time, and leave with a group.


I don't know if this is the answer you are looking for, but generally start by asking "what would I do differently"?

So, if I am running a business (to apply some context), at how often do I not have coverage for certain positions. How can I schedule shifts (for example) to ensure that all critical positions are covered but I have as few people in the building as possible (minimising cost)?

As a simple time series, you could apply any sort of forecasting to this data set - predicting who and how long they will be in the building, when they will arrive next, for example. Look for seasonality in the arrival patterns.

You could also look to predict when a given person will next be in the building. And can you detect outliers? Who has been leaving early or breaking from their usual pattern of behaviour?


I am assuming the context is office setting.

Study of an Impact of a systemic event. e.g. how an organizational event impacted the attendance, or reverse of it (what happened which caused a shift in pattern).

Build employee profile based on localized events e.g. school vacations, time-off patterns, response to extra workload. May not necessarily be "fun" (your question), but profiling can also be done on demographics (age, gender, ethnicity, etc.)

Department or team analysis. e.g. generate networking index of a person, team or department (how correlated are their activities).

Possible impact on internal org design (e.g. scale/reduce support staff services, resource waste reduction, etc. ), use occupancy rate of property to decide on hotel office model, or relocation.

  • 1
    $\begingroup$ btw - a million entries for 50 person sounds like you have many key card doors in this office or somehow the job requires that. Approx 25 entry/exit pairs per person per day for 365 days of year. $\endgroup$ – S2L May 31 '15 at 11:53

Not the answer you're looking for? Browse other questions tagged or ask your own question.