Choosing the correct evaluation metrics is a key step in any AI project. Comparing different models, or evaluating the effect of changing model parameters is only possible with a solid grounding in performance measurement.
Hit '>Play' to watch Lucidate's Evaluation Metrics Video
The lexicon of performance measures can be intimidating: Accuracy, precision, specificity, sensitivity, false positive rates, error, receiver-operator-characteristic. All have a distinct meaning and purpose. This short video will describe why these measures exist, what they are used for and how they are calculated.
Many of the terms and techniques date back to the early days of RADAR. Radar systems are classifiers. They determine if an object of interest is there, or if it isn't. They will exhibit behaviour incorrect classifications. False Positives when they assert that something is there, when it is not. As well as False Negatives - missing a key object of interest. Much of the measurement science & mathematics of modern AI systems has its origins in 1940s RADAR systems.
After watching this video you will understand what all these key metrics mean and how to apply them to your AI and ML projects.
