| Description: | This data comes from a Kaggle competition in 2014 (see link). The problem is to determine whether the subject has been shown the picture of a face or a scrambled image based on MEG, independent of the subject. This data set is just the training data from the competition, split by patient into 10 train subjects (subject01 to subject10) and 6 test subjects (subject11 to 16). This data cannot sensibly be randomly resampled. A leave one patient out approach is the most suitable.
There are about 580-590 trials per patient, giving 5890 train trials and 3524 test subjects.
Each trial consists of 1.5 seconds of MEG recording (starting 0.5sec before the stimulus starts) and the related class label, Face (class 1) or Scramble (class 0). The data were down-sampled to 250Hz and high-pass filtered at 1Hz, giving 62 observations per channel. 306 timeseries were recorded, one for each of the 306 channels, for each trial.
|
