Toy spectral datasets that ship with the package 
from fastcore.test import *

source

load_toy_mir

def load_toy_mir():

Load the toy MIR dataset.

Exported source 
def load_toy_mir():
    "Load the toy MIR dataset."
    df_mir = pd.read_csv(files('soilspectfm')/'data'/'toy_mir.csv.gz')
    ws = df_mir.columns.astype(int).to_numpy()
    X = df_mir.values
    return X, ws

load_toy_mir returns 50 MIR spectra as X, with shape (50, 1701), and their wavenumbers ws, from 600 to 4000 cm⁻¹ every 2 cm⁻¹. The data ships with the package, so it loads without a network connection:

X, ws = load_toy_mir()
test_eq(X.shape, (50, 1701))
ws[[0, -1]]
array([ 600, 4000])

source

load_toy_noisy_mir

def load_toy_noisy_mir():

Load the toy noisy MIR dataset: spectra, wavenumbers and sample IDs

Exported source 
def load_toy_noisy_mir():
    "Load the toy noisy MIR dataset: spectra, wavenumbers and sample IDs"
    df = pd.read_csv(files('soilspectfm')/'data'/'toy_noisy_mir.csv.gz')
    sample_ids = df.sample_id.values
    wavenumbers = df.columns[1:].to_numpy().astype(float)
    spectra = df.iloc[:, 1:].to_numpy()
    return spectra, wavenumbers, sample_ids

load_toy_noisy_mir returns 48 noisy MIR spectra, with shape (48, 3315), their wavenumbers from 599.9 to 3998.4 cm⁻¹ (about every 1 cm⁻¹), and a sample ID for each spectrum. The scans come from a potassium (K) spiking experiment, for a publication in progress:

X, wns, smp_id = load_toy_noisy_mir()
test_eq(X.shape, (48, 3315))
smp_id[:3]
<StringArray>
['TM4.1-0-2', 'SPA1-3-2', 'SPA1-2-2']
Length: 3, dtype: str