An organization is creating a predictive model. During data collection, hidden entries are introduced that degrade accuracy. Which measure best addresses this situation?
Assign model validation to a team member with relevant expertise.
Perform data verification and anomaly detection prior to model training to detect manipulated inputs.
Exclude identified records that are outdated or incomplete from model training.
Disconnect the dataset from the network to reduce unauthorized access after collection.
Data verification and anomaly detection help uncover subtle manipulations in training datasets before they can affect model accuracy. This process involves scanning for patterns, inconsistencies, or unexpected values that may indicate tampering. Simply assigning validation to one person, removing outdated entries, or isolating the dataset does not address hidden data poisoning. Proactively checking for anomalies strengthens the model's integrity and ensures cleaner, more trustworthy inputs for training.
Ask Bash
Bash is our AI bot, trained to help you pass your exam. AI Generated Content may display inaccurate information, always double-check anything important.
What is data verification and how does it help with model accuracy?
Open an interactive chat with Bash
What is anomaly detection, and why is it important in predictive modeling?
Open an interactive chat with Bash
What could happen if hidden manipulations are not detected during data collection?