AI Solution Development Methodologies and Lifecycle AAIA Practice Questions (Page 1)

Questions 1–5

1. 1application · medium

   A bank is developing a fraud detection model. The data science team discovers that the dataset contains duplicate records and inconsistent date formats. They also notice that the minority class (fraudulent transactions) is severely underrepresented. What should the team do to prepare the data for model training?

   • AStandardize date formats and increase the model's complexity to handle the imbalance
   • BRemove duplicates and train the model on the original data to avoid introducing bias
   • CUse only the fraudulent transactions to train the model to ensure focus on fraud
   • DRemove duplicates, standardize date formats, and apply a resampling technique to balance the classes
   Reveal Solution Discussion Select an answer first
2. 2expert · hard

   An insurance company has deployed a model to predict claim fraud. The model was trained on data from the past five years. After one year in production, the company notices that the model's precision has decreased, but recall has remained stable. The data science team suspects that the model is now flagging too many false positives. What is the most likely cause of this change?

   • AThe threshold for flagging fraud was set too low during deployment
   • BThe distribution of fraudulent claims has changed, causing the model to misclassify legitimate claims as fraudulent
   • CThe model has become overfit to the training data over time
   • DThe model is not being retrained frequently enough
   Reveal Solution Discussion Select an answer first
4. 3application · medium

   A retail company wants to build a recommendation engine to increase cross-selling. The data science team has access to historical purchase data, but the data is stored across multiple legacy systems with inconsistent formats and missing values. The company has a strict budget and cannot afford extensive data engineering. Which feasibility assessment should the team conduct first?

   • AOrganizational feasibility of training staff on the new system
   • BTechnical feasibility of integrating the legacy systems
   • CRegulatory feasibility of using purchase data for recommendations
   • DData feasibility, including quality, completeness, and accessibility of the historical purchase data
   Reveal Solution Discussion Select an answer first
5. 4foundation · easy

   An organization plans to develop an AI solution where requirements are expected to evolve as the project progresses and early delivery of a minimal viable model is desired. Which development methodology is most appropriate for this scenario?

   • AWaterfall
   • BDevOps
   • CCRISP-DM
   • DAgile
   Reveal Solution Discussion Select an answer first
6. 5foundation · easy

   Which of the following is an ethical consideration that should be addressed during the AI lifecycle?

   • AEnsuring the model is deployed on the latest hardware
   • BEnsuring the model does not perpetuate or amplify biases
   • CEnsuring the model is cost-effective to run
   • DEnsuring the model is trained using a specific programming language
   Reveal Solution Discussion Select an answer first