Classifier question

Just a quick question when using the classifier:
Does having copies of a cell in one class affect the rules? I’ve been manually deleting any duplicates in any class but perhaps this is unnecessary. Maybe I want to have duplicates to “tell” the classifier that it made a good decision using its classification rules.

We generally tell people not to worry about duplicates. If you classify the same cell as positive twice, then that set of measurements will contribute twice to that class. Likewise, if you classify the same cell as positive and negative (which no one will admit is possible, but most likely everyone will do), this is essentially telling classifier that this particular set of measurements is not terribly good at differentiating your classes. These are both good things as long as you’re creating a reasonably big training set (and if you’re not creating a big enough training set, then you should be training more :wink: ).