How can you identify research gaps in transportation systems analysis?

The literature is important, as are the other key elements identified here, particularly thinking critically. For those of us in local planning and transportation systems, if your time is limited, you need not look into studies for every type of community, but rather focus on those which reflect areas like your own. You can always do more later. Thinking and reviewing critically what you find at the suitable nature and scale of local systems will help to focus your research (i.e., rural, small town and urban all have their own needs and opportunities). From there, think forward to what your community needs not just now, but in the future.

Transportation systems do serve communities, and communities do live in a real world, not ideal and featuring its own characteristics in terms of land use, urban development etc. Transportation planning should go hand in hand with urban planning and the two levels should be analyzed together. Ideal theoretical systems will not work everywhere and adaptation to local conditions is fundamental. More cooperation between urban planning and transportation professionals is much needed

When back in the day I reviewed the literature for packaging and its lifetime, quality, etc. In the literature many papers published for the corrugated paper is one person and their colleagues that are focused on McKee formula. From there many experimentation conducted to change transportation safety parameter which is for local (3) global by land, sea or air mode except overseas (5) and for overseas (6). But since these values are mere approximations and rounding to the ceiling integer. They can be revalued by correct experimentation techniques

After thorough literature review, decided to work on finding essential parameters that change the quality factors and health of the box factors(After review it is possible to see that these factors are dependent to each other) and these dependent values also change by other main characteristics of the environment and the quality of the design. In which you can change the design when their eigenvalues are determined and compared with each other. Such as in finding likelihood estimators, here we find our best predictors and adjusting them for cheaper cost without reducing the quality when wetted or so.

These data we acquired from the designed data acquisition system after domain expert knowledge acquired and now that we know the theoretical parameters that affect the parameter that we want to change, we can use Unsupervised learning techniques after EDA. After we see the linear and non-linear relations we can look for eigenvalue decomposition of the parameters find their coefficients and their power estimators linearly or by nonlinear methods(tSNE or so) a group of electronical engineer, signal processing expert and data scientist can find in which mode of transport what effect cause more variation on these factors and how to effectively manipulate these for our benefit

Later the given results, one can build a supervised model on the failures and passes from the returned delivery and make a predictor for the factors to be changed after setting the thresholds from learned variability and characteristics that determine the health of the delivery throughout the process. By that you can make an automated ML pipeline and give inputs for the specified parameters(geometry, corrugation shape, material, wetness, RH% or so) and get a classifier result as fail or pass for the safety/health parameter. This way you can connect theory and reduce errors and variations that are played significant role in the past and led to creation of expensive packaging for reduced return or low quality

Many other experimentations can be done. Also for data acquisition, it is important to consider Nyquist-Shannon limit, other variables that may lead to changes which are not discussed in the literature, experiment as much as possible for reliability measures. Consider designing the experiments by OFAT, factorial design and other techniques as much as possible