The script for scoring the submitted algorithms is score.m.

The organizing committee selected 300 records randomly from the training set as validation data (including the corresponding REFERENCE.csv). storing in the validation_set folder. The validation dataset is only to verify every challenger's code running successfully.

The score.m requires that participants downloaded the validation data ('validation_set' file) in the directory where score.m is located.

In score.m, challengers need to fill in their names (Participant_Name);
                                Data file path (data_file_path)
                                REFERENCE.csv path (ref_file)
For example,
Participant_Name={'A'};               
record_path= D:\cpsc2018_matlab\validation_set\;
ref_file=D:\cpsc2018_matlab\validation_set\REFERENCE.csv';

The frame of cpsc2018.m is provided, challengers can write their algorithms in to this function.
classify_result=cpsc2018 (record_path)

INPUTS:  data_file_path, ECG signal data file path
OUTPUTS: classify_result, it is table format,including two columns: the first column is recording name; the second column is prediction label.
for example: Recording    Result
   A0001         1
                               .           .

The classify_result will return to the score.m function. After all ECG data were classified, the score (including F11, F12, F13, F14, F15, F16, F17, F18, F19, F1, Faf, Fblock, Fpc, Fst) was given.  

Two files should be produced in the local path: the Reslut.csv in charge of saving the inference results corresponded with record file and the score.txt storing the final scores.

It is hoped that each team can optimize the time complexity of their algorithm while running code under validation dataset, and submit the algorithm run time on classifying 300 recording files of the validation dataset as a txt file named time_validation.txt based on the local platform. We will predict the execution time of the algorithm on the test dataset based on the submission and schedule the follow-up evaluation cycle:
If the algorithm runtime on the test dataset is less than 30 mins, we will give scoring feedback in 3 days;
If the algorithm runtime is between 30 and 120 mins, we will give scoring feedback in 1 week;
If the algorithm runtime is over 120 mins, we will not guarantee the feedback in time, please understand!
