Statistical learning for data-limited systems

Question

I'm currently conducting a review for quantitative methods being used for tropical inland fisheries. One of the major problems for modeling methods in tropical inland fisheries is the lack of data available. Fisheries assessments are difficult with widely distributed, small-scale fisheries. As many of the people living in tropical regions are subsistence fishers, they directly consume fish without any recordings of the catch.

I'm trying to find statistical/mathematical modeling methods that are able to deal with data-limited systems. I do not have a significant background in statistics, but I am curious about the concept of "statistical learning" and data science methods in general. My understanding of current "hot" topics in data science (machine learning, AI) is that they are useful for big data. I have not come across their usefulness in data-limited methods. Although it seems as though they could potentially deal with data limitation by using a range of other data available to make hypotheses about the state of the data that is limited. Do you know of any usefulness of statistical learning or data science methods in general for data-limited systems? Thanks!

Answer 1

It really depends on the data set in hand. For example for short time series you can look at this thread.

In general I would say that non-parametric statistics works pretty well for smaller datasets.

Answer 2

The question is - what are you trying to answer. This will dictate how you approach the problem. Are you interested in tropical fish lifespans? Survival analysis would be your answer which has methods for dealing with data limitations. Other objectives will necessitate other methods