Here are a few announcements:

Regarding the 1st project, I would like to emphasise that when presenting the classification results it is way more important to present a confusion matrix containing the number of correctly and incorrectly classified cases (absolute values), than a normalized confusion matrix (with values between 0 and 1).

Knowing the absolute number of cases is particularly important when the classes are imbalanced, and helps to explain why you might be getting "inconsistent" results.

J:Paulo