4. A person wanted to use an AI model to predict the price movement of stocks. However the person chose to use data only from the largest companies as a few examples even though they wanted to use it to predict stock price movements of smaller companies. What is the type of bias described in the scenario?
Answer: A
Selection bias
The scenario describes selection bias, which occurs when the sample used for analysis does not represent the target population adequately. By focusing solely on data from the largest companies, the individual fails to capture the broader market dynamics that could influence smaller companies' stock price movements.
A) Selection bias
This option is correct because the individual’s choice to use data exclusively from large companies creates a skewed representation, impacting the model's ability to generalize predictions for smaller companies. This misrepresentation leads to inaccurate conclusions about the stock market as a whole.
B) Algorithmic bias
This option is incorrect as algorithmic bias pertains to biases that arise from the algorithms used for data processing or decision-making. In this case, the issue is not about the algorithm itself but rather the selection of data points that do not represent the intended population.
C) Confirmation bias
This option is incorrect because confirmation bias involves favoring information that confirms existing beliefs or hypotheses while disregarding contradictory data. Here, the focus is on the selection of a specific dataset rather than an individual's inclination to validate preconceived notions.
D) Measurement bias
This option is incorrect as measurement bias refers to errors that occur during data collection or measurement processes that affect the accuracy of the data. The scenario does not highlight any issues with how data is measured but rather focuses on the selection of the data used for predictions.
Conclusion
Selection bias is the definitive issue in this scenario, as it illustrates how the choice of a non-representative sample can lead to flawed predictions. All other options fail to accurately describe the core problem, which lies in the inadequate representation of the target population necessary for effective stock price movement predictions.