Ecosphere - Let’s Do the Time Warp Again: Non-linear time series matching as a tool for sequentially structured data in ecology
Description
These CSV file include all of the raw data used in the article "Let’s Do the Time Warp Again: Non-linear time series matching as a tool for sequentially structured data in ecology". File: juvenile_Sr_isotope_transects_Mendeley_data.csv This file contains the strontium isotope data for each otolith in wide format. The column information is as follows: unique_otoID - unique sample ID mean - mean of 87Sr/86Sr over entire transect sd - standard deviation of 87Sr/86Sr over entire transect median - median of 87Sr/86Sr over entire transect natal_downstream - known capture location and downstream capture location. Format; First capture location --> Second Capture Location columns 0 through 1152.8 - This is the number of microns from the otolith core, values are standard corrected 87Sr/886Sr File: juvenile_element_transects_Mendeley_data.csv This file contains the trace element data for each of the otolith samples in the study. Each row represents one interpolation on the ICPMS. This is in "long" format, the opposite of the strontium isotope ratio data. The column information is as follows: Order - This is the original order of the data. Time (and thus microns from the otolith core) goes from lowest to highest number within each otolith. Microns should be interpolated based on the microns from the "master" distance in the 87Sr/86Sr data. OtoID - unique sample ID, matches that of the strontium isotope data Sr_Ca - standard corrected strontium to calcium ratio in millimoles/mole Ba_Ca - standard corrected barium to calcium ratio in millimoles/mole Mn_Ca - standard corrected manganese to calcium ratio in millimoles/mole
Files
Steps to reproduce
Refer to the published manuscript for specific methods. The easiest way to deal with this data for analysis in {dtwclust} is to combine the element data with the strontium isotope data into a list of data frames containing the data only, with each list named using the unique sample ID. These dataframes can be fed directly into the tsclust() method. Doing this requires reinterpolating both sets of data (200 cells was used in the paper) and then combining them.