This part, contains project sample for Classification in Python using Sci-Kit tools.
In order to run the project please follow this link.
There are two classes in this dataset, The red wine class and white wine class. These data has been combined together and then, using a Classification method, we try to separate them. The following method has been applied on the data and the Result can be viewed as below. Obviously, Linear Discriminant Analysis (LDA) worked best on these data.
| Method | Data | RSS | TSS | R2 | Precision | Recall | F-Score |
|---|---|---|---|---|---|---|---|
| Logistic Regression | Train Error | 90.00 | 832.964 | 0.892 | 0.970 | 0.976 | 0.973 |
| Test Error | 32.00 | 359.303 | 0.911 | 0.976 | 0.979 | 0.978 | |
| Linear Discriminant Analysis | Train Error | 26.00 | 843.111 | 0.969 | 0.992 | 0.993 | 0.992 |
| Test Error | 7.00 | 359.812 | 0.981 | 0.994 | 0.996 | 0.995 | |
| Quadratic Discriminant Analysis | Train Error | 67.00 | 864.928 | 0.923 | 0.986 | 0.975 | 0.980 |
| Test Error | 20.00 | 367.443 | 0.946 | 0.990 | 0.982 | 0.986 | |
| Gaussian Naive Bayes | Train Error | 140.0 | 876.597 | 0.840 | 0.968 | 0.951 | 0.959 |
| Test Error | 45.00 | 374.056 | 0.880 | 0.978 | 0.962 | 0.969 | |
| Linear Regression | Train Error | 120.097 | 724.029 | 0.834 | |||
| Test Error | 45.993 | 316.009 | 0.854 |
Here is the ROC Curve of the mentioned methods using 5-Fold Cross Validation:
Here is the ROC Curve of the mentioned methods using Leave-One-Out Cross Validation:
