Triton is the second largest shipping container company in the world, with over 200,000 containers distributed in ports around the world when I was retained to perform a statistical analysis. The company has since been acquired by Warburg Pincus and Vestar Capital Partners.
There were several problems to analyze, including: price models based on container location, size, and history of demand; retiring older containers; and the project I specifically worked on, which was container idle time. The problem at hand was that when containers sit idle in port, there is an implicit cost. The study I carried out was to determine, within the constraint of the price model (which dictates to a large degree when containers will go from idle to active), if the idle times that containers were accumulating was better, worse, or as would be expected.
This project took approximately four months to complete, as I gathered and assimilated container data from ports around the world: Portugal, Egypt, China, etc. The tricky part was building a statistical model which took into account the key elements of the pricing algorithm, and also to a certain extent adjusting to a sometimes moving price model. I recall about 1 month into the project talking to my team lead, who basically told me I'd likely have to start over based on a new dimension to the model that he had left out. That was fine, as I was able to incorporate the new information he provided and continue on.