There we were. Three against a hundred.
Toughest three guys we ever ran into.
The growing field of data science has a quiet undercurrent that the scientific approach and the use of hard facts will yield understandable results. There is one problem with that belief. We still need good old, easy to misinterpret language to communicate with each other.
As per the example above, the last line was probably not where you thought the first two lines were going. Problems or opportunities will be articulated to the analysts and data scientists charged with finding an answer. They in turn will communicate their findings back to the requester. All along, the pesky ease of having a slightly different understanding of the terms in use will plague the process.
If someone was tasked with quantifying the enormity of customer service interactions for a company, many analysts would develop models to classify interactions and different means to measure the size of the transactions. Catch is, enormity is not a variation of enormous as many believe. It is specifically associated with something bad that someone has done. Or in this case, the topic of interest is customer service interactions that went badly because of the company.
There is much discussion about data quality and its negative impact on analysis. Keep in mind as well that the very first question asked will insert uncertainty and complexity into any analytical activity.