Prepare for the IBM Data Science Exam. Utilize flashcards and multiple-choice questions with hints and explanations to hone your skills. Get exam-ready now!

Practice this question and more.


Which is a potential drawback of supervised learning?

  1. It requires vast amounts of data.

  2. Labeling the data is arduous and expensive.

  3. They are not used much as of late.

  4. Clustering is difficult in supervised learning.

The correct answer is: Labeling the data is arduous and expensive.

The potential drawback of supervised learning relates to the process of labeling data, which can indeed be arduous and expensive. In supervised learning, the model learns from a labeled dataset, meaning that each data point is paired with an output label that indicates the desired result. This requirement can lead to significant challenges, especially in domains where gathering accurate labels is time-consuming or requires expert knowledge. For example, in fields like medical imaging or natural language processing, having skilled professionals label the data can incur high costs and take considerable time. If the labeling process is not managed efficiently, it can hinder the training of the model, resulting in delays and potentially limiting the amount of data that can be realistically used. While vast amounts of data are indeed often required (as mentioned in another choice), it’s the labeling aspect that entails a direct, ongoing resource investment, which can be particularly challenging for organizations. The other options do not accurately capture the nuances of supervised learning challenges, making the issue of labeling a significant drawback.