4 ms·
You know since this assignment is basically a boss asking for a single report once about "loyal customers" complexity doesn't matter. let's say worst case scena
by interactivecode 3y ago
You know since this assignment is basically a boss asking for a single report once about "loyal customers" complexity doesn't matter. let's say worst case scenario it takes 30min to run. who cares about complexity the business value is in the answers from the data.
If you're consistently getting reliable answers and finally decide to build a system for these types of reports, clearly this guy's real world experience at Amazon's Clickstream product is going to be far more valuable than what ever anyone who is brand new to the problem can come up with, even if they choose the "correct" algorithm from the start.
Because I bet you that for most real world products that create more than a single fixed format report you actually want your data setup in a completely different way than what you initially thought. You'll probably learn for example that you want to aggregrate data per week instead of per day. or perhaps you want to link this data to an internal users database, or perhaps your boss wants a notification when new data is added. Or perhaps you'll learn that loading it into a single 1GB SQLite DB solves your problem without even needing to think about any algos.
- extragood 3y agoI have a similar feeling. My team builds out a lot of one-off projects for customers, and a lot of them don't need to be especially performant. That has always been the case for reporting/analytical projects. You know what does make a difference though? Amount of development effort spent on the solution. The most performant solution can very easily cost the company much more in terms of development time spent. So in that sense, it's sub-optimal