dataset-suite documentation

A general data handling toolkit with a powerful class for abstracting away work with multidimensional set.

Parts of this code were created as part of PhD work supported by the EU ITN EID project CORAL (GA no. 859841).

Installation

You can install this package from pip:

pip install dataset-suite

Once installed, all useful functions are in the top-level module:

import dataset_suite as ds

Basic usage

A short introduction to using the library:

import dataset_suite as ds

# Create a dataset out of a set of spectra
data = ds.dataset(
    array,
    index=np.arange(array.shape[0]) # A basic numerical index
    wavelength=spectrometer.wavelength # You can store a numpy array as an axis descriptor
    # alongside your data. This way you know that the second axis of this dataset is a
    # "wavelength", and what wavelength corresponds to each datapoint
)
data.metadata["comment"] = "afternoon spectra"
# Save the dataset to a HDF5 file
data.save_h5("file.h5")
# Load a dataset from a file
data = ds.load_h5('file.h5')
# inspect the dataset - what axes does it have?
print(data)
# dataset(index[10], wavelength[100])

data.raw # access the underlying numpy array
data.wl # use axis names as attributes to get the values for a given axis,
#         like wavelength here
data.wl[10] # this is the wavelength corresponding to the 10th column of the array
data.take(index=5) # slice along the index axis: dataset(wavelength[100])
data.take_sum("wavelength") # sum along the wavelength axis: dataset(index[10])
data.metadata # dictionary of measurement metadata

The library also includes some convenient helpers like a function mapping values to colours using matplotlib colormaps, and a number of filename search/analysis tools. Have a look at https://dataset-suite.readthedocs.io/en/latest/api.html

Contents:

Indices and tables