Confusion Matrix Made Simple



Confusion Matrix Made Simple

When I started with Data Mining 101, the first thing I came across was known as “The Confusion Matrix”. It seemed interesting at first so I started diving deep. It looked very simple at first, but then when more terms were introduced, things become complicated rather confusing; and then I realized the reason it is called “The Confusion Matrix”.

The terms True/False Positives/Negatives are really very confusing to remember and hence I found a sleek way to understand it.

Just add “ly predicted as” between two words

So, True Positive → Truly predicted as Positive False Positive → Falsely predicted as Positive True Negative→ Truly predicted as Negative False Negative → Falsely predicted as Negative

Interpreting True/False Positive/Negative

True Positive: The data point is a True Positive if it is truly predicted as positive. If I try to elaborate truly we get something like this: The data point is a True Positive if it is actually positive and is truly predicted as positive.

For a food lover:

When a Burger is predicted as Burger and not Hotdog it comes under True Positive.

  • Burger - Positive Class
  • Hotdog - Negative Class

For an animal lover:

When a Dog is predicted as God and not Cat it comes under True Positive.

  • Dog - Positive Class
  • Cat - Negative Class

False Positive: It will be interesting when we try to interpret False Positive. The data point is a False Positive if it is falsely predicted as positive. If I try to elaborate falsely we get something like this: The data point is False Positive if it is actually Negative but it is falsely predicted as positive.

Again, for a food lover:

When a Hotdog is predicted as Burger, it comes under False Positive

Again, for an animal lover

When a Cat is predicted as Dog, it comes under False Pos-itive

Rest you can derive on your own 😉

Conclusion

In conclusion I would like to say … add “ly predicted as” in the middle.