Missing Data & ML Models
What are the three types of missing data?
MNAR, MCAR, MAR
What is it called when a model is too complex?
Overfitting.
What type of software interface allows different applications to communicate and integrate services?
APIs.
This term refers to the difference between a model's predicted value and the actual observed value.
Error
This term refers to understanding and explaining how a machine learning model makes its predictions.
Interpretability.
This machine learning model splits data into branches based on feature values, forming a structure similar to a flowchart.
Decision Tree
What technique prevents overfitting?
Regularization.
What Python library is widely used for data analysis?
Pandas.
This type of error refers to the difference between a model's prediction and the true value due to the model's assumptions.
Bias
These are the conditions that must be met for a model, like linear regression, to produce unbiased and reliable estimates. Do not violate them.
Model Assumptions.
What is one common method for handling missing data?
Imputation
This correlation measure is used to assess the strength and direction of the monotonic relationship between two variables, based on their ranks.
Spearman Rank Correlation
What process involves identifying and resolving errors in software to ensure correct functionality?
Debugging.
This type of error refers to the model’s sensitivity to small fluctuations in the training data.
This interpretability method is based on Shapley values from game theory and provides consistent, global explanations for feature importance.
SHAP
This regression technique extends linear regression by adding higher-degree terms to model nonlinear relationships.
Polynomial Regression
What process involves selecting a subset of individuals from a population to draw conclusions about the entire group?
Sampling Methods.
What AI research lab, originating from China, focuses on open-source large language models?
Deepseek.
This concept describes the tradeoff between two types of errors.
Bias Variance Trade-off
What term describes machine learning models whose decision-making processes are not easily interpretable?
Black-box models.
A non-parametric model that classifies new data points based on the majority class of their nearest neighbors.
KNN
What statistical technique determines the minimum sample size required for detecting an effect?
Power Analysis.
Developed by Meta, this family of open-weight large language models is named after a South American animal.
LLaMA.
What statistical technique is used to ensure models generalize well by preventing overfitting?
Regularization
In time series analysis, this term refers to when past values are used as predictors for future values, introducing delays.
Lagging.