Whenever you're ready, walk me through how you'd approach this.
Describe how you would use a confusion matrix to evaluate a binary classifier in a Careem data science project, and what metrics you would derive from it to guide model improvements.
Explain the roles of true positives, true negatives, false positives, and false negatives. Derive appropriate metrics, discuss their trade-offs, and describe how the findings would influence threshold selection and error analysis.