Hyperspectral data for different glucose and silicon concentration

The whole dataset contains 94 730 samples of all glucose concentrations, and for silicon, there are a total of 47 375 samples. The obtained data was initially processed toclassify its quality and explore the potential for predicting data based on it. The whole dataset hyperspectral data for all glucose concentration is equal. The dataset has:

Spectral data range from 892.958 [nm] to 2506.03 [nm] in intervals of approximately 9 nm between each channel.

Spectral data is stored in the int64 format.

It has 94730 samples for each wavelength.

Skewness coefficient for the entire wavelength range is positive and ranges from 0.2 to 8.5, excluding the wavelength 902.503 [nm].

Kurtosis for wavelengths in the range of 902.503 - 1312.93 [nm] and 2315.13 - 2506.03 [nm] has values from -0.2 to -1.7, while for wavelengths 1389.29 - 1580.18 [nm], it ranges from 70 to 163. For wavelengths 1608.82 - 2019.24 [nm], this coefficient is between 25 and 55, and for the remaining wavelengths, it ranges from 0.2 to 5.

The linear correlation coefficient between wavelengths and glucose concentration using the Pearson method is approximately -0.3.

There is a readme file with a description of the dataset in the folder with measurement data.