January 7, 2021
We know that for unbalanced datasets we can’t rely on accuracy. Here we have to measure the performance of the model using Precision & Recall. The ideal scenario would be where both Precision & Recall are high. But research has shown that the two metrics are somewhat inversely proportional to each other. I said somewhat because, Precision does decrease with increase in Recall, but beyond a certain value of Recall, Precision value becomes constant i.e. indifferent to change in recall. Following graph demonstrates the relationship beautifully:
by : Monis Khan
Quick Summary:
We know that for unbalanced datasets we can’t rely on accuracy. Here we have to measure the performance of the model using Precision & Recall. The ideal scenario would be where both Precision & Recall are high. But research has shown that the two metrics are somewhat inversely proportional to each other. I said somewhat […]