ITU-ML-Final_Project

Pretty much: use a very small dataset of refraction index and chemical composition of different glass shards and create a couple classification models and try to do as well as possible.

Classification

We need to explore at least five classification methods:

M1. Linear or quadratic discriminant analysis as you see fit. (Casper)

M2. Decision Trees (Tijn)

M3. Support Vector Machines (Flavia)

M4. k-nearest neighbours using 2 features that we have chosen by dimensionality reduction. (Flavia)

M5. One or more classification methods of our own choice. (Whomever idk)

The first two methods, M1 and M2, should be implemented entirely by us. For these two methods, we may use only any standard Python libraries and the numerical libraries NumPy and SciPy. The only machine learning library we may make use of is TensorFlow, i.e Keras, or any other similarly high-level APIs are not allowed.

For the remaining methods, M3, M4, and M5, we may use any library we wish with no restrictions.