Column
Exploring the wine dataset: A bottle of wine was contaminated with arsenic. The task is to build a model that will tell us which vineyard (i.e.Class) the wine with the arsenic came from and, therefore, the “guilty party”. Source of the data: HDclassif package
Column
Classification using Linear Discriminant Analysis (LDA) and Quadratic Discriminant Analysis (QDA). Comparing the performance of the cross-validated LDA and QDA models, it turns out that the LDA model’s accuracy is 98.9% whereas QDA classified correctly 99.21% of the data. This means that quadratic discriminant analysis performs slightly better than LDA. Once the chemical analysis of the poisoned wine is in, we use our QDA model to predict which vineyard it came from. The model predicts that the poisoned bottle came from vineyard 1. Time to go and take necessary measures. Chart: LDA classification of the 3 vineyards.
