Member-only story
Confusion matrix 🤔
Hello Readers!!!
This term isn’t so named because it’s hard to understand!
In fact, a confusion matrix makes it easier to understand the results of your algorithm. It’s named because this visualization makes it clear whether the algorithm is mislabeling (or confusing) one class with another.
Reading a confusion matrix, however, does take a little practice. For each variable you’re comparing, the matrix will list how many of that variable have been classified or misclassified according to each column’s variable in the row for that variable.
Then, in each column, it’ll list how many of each row’s variable has been classified or misclassified as the variable in that column.
Producing a confusion matrix for your supervised model will give you a better idea of how accurate it is — in other words, how many variables it’s misclassified.
Why do we need a confusion matrix?
The primary reason is to assess the performance of classification models. These models, which include spam filters for your email or facial recognition on your smartphone, need to accurately categorize data into distinct classes. However, no model is perfect; mistakes happen.
