Included are all the R programming scripts that I have used towards my dissertation, and more that helped me get there.
The 'and more' part, involve word documents of note-taking to help me better understand and inspire myself into understanding what I needed as a next step in programming the right function for my objective.
This dissertation includes familiarity with machine learning, calibration models, linear regression, and covariate analysis with eBird citizen science data.
Abstract:
There have been declines in the Red-headed Woodpecker since the last 50 years, and these implications are related towards declines in Savannas, Snags and dead trees that are effectively limiting the species distribution range. To understand these declines the study involved multiple BCR regions across North America along with multiple landcover classes to understand if the declines persist, and whether its range-dependent, or even landcover class dependent. Furthermore, the use of modern analytical techniques through machine learning and calibration models were implemented to determine recent trends. The use of recent Landsat data from Modis MCD12Q1, provide accurate estimates towards these trends. Given that, no research has yet been appended towards the woodpecker in grassland and croplands, the results were surprising and have shown high probabilities for encounters in these areas. However, it was also found that several BCR regions are continually declining and in need of further research and conservation to understand what the indicators for decline really represent. Recent literature has shown it is related to short burn-intervals, agricultural intensification, or even urbanisation.
Key words: Probability, Encounters, Landcover changes, Distribution models.