| name | csv_npy |
| description | Plain .csv / .npy / .npz array loaders — last-resort generic data import |
| layer | L0 |
| group | consumer_eeg |
| metadata | {"tags":["io","csv","npy","npz","generic","scientific_python"],"formats":[".csv",".npy",".npz"],"modalities":["eeg","meg","seeg","ecog","fnirs","lfp"]} |
CSV / NPY / NPZ
Format Overview
Plain scientific-Python arrays:
.csv: text, row-major or column-major.
.npy: numpy save format; one array.
.npz: numpy archive; multiple arrays.
Used as last resort when no domain format applies, or as an
intermediate between pipeline steps.
Loader
Primary:
np.load for .npy / .npz.
pandas.read_csv or np.genfromtxt for .csv.
Common Pitfalls
- Orientation.
(n_channels, n_times) is EasyBCI convention; CSV
often is the transpose.
- No metadata. Sample rate, channel names, units — must come from
a sidecar or user declaration.
- Memory mode. Large
.npy should use mmap_mode="r".
EasyBCI Path
- load:csv,sfreq=USER_PROVIDED,layout=channel_x_time
- load:npy,sfreq=USER_PROVIDED
References
- (none specific — generic format.)
Boundary
mat: MATLAB array variant.
custom_binary: raw binary without numpy framing.