A manufacturing company inspects all products before selling them. Less than 1% are defective and do not pass inspection. You are developing an algorithm to predict defects before product inspection. Correctly predicting defects at least 5 days before inspection significantly reduces costs. Conversely, false positives must be avoided as they result in added costs. The manufacturer wants to address as many defects as possible. Which problem statement best addresses the description above? Select the best answer.
a. We have a large dataset of product inspection records and want to understand what products are likely to be defective.
b. We want to predict product defects five days before inspections occur to reduce costs. The algorithm must have high accuracy (>95%) and high recall (>95%), even if that means low precision
c. We want to predict product defects before inspection to reduce costs. Furthermore, we want to identify why products are defective so we can take the right corrective action.
d. We want to predict defects five days before inspections to reduce costs. The algorithm must have high precision (>95%). Higher rates of recall lead to more savings only if precision is maintained above 95%.