This folder contains scripts for the classification of thick-billed murre (TBMU) behaviour from acceleromter data. Details of data files and classification methods are provided in Pattterson et al - A comparison of techniques for classifying behaviour from accelerometers for two species of seabird

accFunctions - contains functiosn called in the other scripts and should be saved in the working directory with other files from this download
process_accelerometer_TBMU - this script can be used to calculate accelerometer-derived metrics from raw acceleration data. The data uploaded with this paper have already been processed using this data, but this script was included to show how this was done.
combine_Files_TBMU - the data archived for this ppaper had to be separated into multiple files, this script will combine and save all the TBMU data files in a single .csv
get_Training_Data_TBMU - this script will create a randomly sampled training data set from the TBMU data for use in training random forest and neural networks, this script should be run before using 'classify_RF_TBMU.R' or 'classify_NN_TBMU.R'
classify_HS_TBMU - this script uses a histogram segregation method to classify TBMU behaviour
classify_RF_TBMU -  this script uses a random forest to classify TBMU behaviour
classify_NN_TBMU - this script uses a neural network to classify TBMU behaviour
classify_KM_TBMU - this script uses the k-mean method to classify TBMU behaviour
classify_EM_TBMU - this script uses the expectation maximization method to classify TBMU behaviour
classify_HMM_TBMU - this script uses a hidden markov model to classify TBMU behaviour