In nuclear reactors, it is essential to track and predict the amount and type of plutonium present in order to prevent build-ups that can lead to nuclear meltdowns. When different wavelengths (from 0-1650) are run through 110 separate nitric acid solutions for ten repetitions each the resulting data matrix is 1650 x 1100. To predict the plutonium present and the nitric acid used the dimensions of the matrix must first be reduced. In this paper, we proposed a novel application to reduce the dimensions using a Non-Negative Matrix Factorization. The method currently used to reduce the dimensions is Principal Component Analysis. However, when dealing with a potentially catastrophic element multiple reduction techniques should be implored to check the prediction. In this research, the first attempt to reduce the dimensions of plutonium data using a non-negative matrix factorization approach was utilized. The results proved to be more explainable and often more accurate than principal component analysis showing that if implored instead of PCA, or in addition to PCA, this novel NMF approach can help prevent nuclear meltdowns. Additionally this article is aided with a novel supplemental proof conveying that minimizing reconstruction error is equivalent to maximizing variance in principal component analysis.