In all crayon rail games, an important part of winning is building your track efficiently. Knowing which cities generate the best returns, and which to rarely bother connecting, can help you build an efficient network. The loads supplied and demanded by a city largely determine that city's value to your network.
Using data from over 300 four-player matches, the table below shows the percentage of times a Nippon Rails city was served by the winning player's track. The major cities are at 100% because the rules require their service as a victory condition. As is true on many crayon rail maps, cities that monopolize the supply of a load can be important to victory, in this case Kumamoto (gold and pearls) and Kagoshima (glass).
Small cities can be served by the fewest number of players, so those cities can be important to connect early. The top two small cities are Shimonoseki and Niigata, which fall somewhat low in the rankings compared to the best small cities of most other crayon rail maps. The cities least often associated with victory are the distant ones on Hokkaido. These stats were generated via simulation by the EB Player computer program that can be downloaded from http://www.railgamefans.com/ebp
This makes for interesting reading, but except in a top level tournament its difficult knowing how best to make use of this. I don't picture myself connecting to a city on principle if my runs don't happen to lead me there.
Occasionally I may build through a city I don't need on my way to somewhere else, anticipating I may use that city in future so that it's worth the marginal additional expense ... I suppose this chart could indicate the worthwhileness of that philosophy in each instance.
I use the chart's data for both network build purposes and trip planning. I'll build to or near the popular cities early so as to secure a good/best route.