Cleaning and compiling various work on trait correlations across species of UK flora.
Script for pipeline is in src/. The script main.R will scrape all necessary data from the ecoflora, save that data, determine fertilization mode for that data (written to a new column myFert3), then finally save that dataframe including myFert3 to a different file.
Datasets can be found in data/. These include old data, including:
- Old raw data scraped from the ecoflora in June 2020.
- That old scrape data which was processed to determine myFert3 using an old Python script.
I am trying to reproduce the scrape data and determine characteristics of that data functionality of old Python scripts written in the summer of 2020.
I am not able to test this for the scrape aspect, since the website from which the data was scraped has changed in the intervening time. However I can test that the Python scripts to determine myFert3 are reproduced. In sketch_test/check_old_vs_new_data_and_methods.R I test two things:
- The new R script for determining
myFert3acts identically to the old Python script on the old scrape data. - Similarly for the new scrape data.
These tests were passed, but in doing so, a couple of bugs and areas for obvious improvement were identified in the old Python code. At the moment these have been reproduced in the R scripts.
The spread of myFert3 has changed quite significantly:
| Fert Mode | n (June 2020) | n (June 2023) |
|---|---|---|
| selfing | 60 | 57 |
| mixed | 787 | 643 |
| outcrossing | 235 | 185 |
| NA | 2192 | 2390 |
The important aspects of determining myFert3 are in the file src/determine_fertilization_characteristics.R. The main function of this file is assign_3_group_fertilization(). In this file, characteristics of species are determined atomically in functions. These characteristics are combined to determine myFert3 in assign_3_group_fertilization().