<aside> ⚡ In our Data Science team, the nature of this calculation is determined by the category of work the project falls under. We assign our projects into 3 categories: Experiments, Models, Investigations
</aside>
Typical duration: 2-3 months
Experiment protocols encompass any tests that are used to compare different versions of a single variable. The most famous type of xp protocol is AB testing, but they are not limited to that. Experiment protocols are split into different crucial phases (design, monitoring, analysis, etc.) that require data science knowledge to be taken care of properly.
Typical duration: 1-4 months
Models are ensembles of hypotheses and algorithms. They ingest data and output data. A typical example of a model is a time series forecast.
Typical duration: 1-3 weeks
Investigations are meant to understand root causes of unexpected events. The events are for instance large gaps between what we aim at or expect and what we achieve. It is always very instructive to understand the gap’s root causes. When the gap yields a large negative impact, it is urgent to investigate: it might be purely cyclical or we might need to fix something and counteract fast.