Setup | Simulated Logistic Regression Example | One-Step Balanced Sampling | Two-Step Rareness-Aware Optimal Subsampling | Automatically Account for Rarity if rareFeature.index = NULL | Balancing the Outcome for Logistic Regression | Objective Weights | Control Options | Non-Binomial Families

Model and Objective | Two Ways ssp.glm() Works | Pilot Estimator | $$\pi_i^ | Choosing criterion | criterion = "optA" | criterion = "optL" | criterion = "LCC" | criterion = "uniform" | Choosing sampling.method | sampling.method = "withReplacement" | sampling.method = "poisson" | Choosing likelihood | likelihood = "weighted" | likelihood = "logOddsCorrection" | $$P(Y_i = 1 \mid x_i, i \in S) | Method Summary | Example Data | Example 1: Default Optimal Subsampling | Example 2: Logistic Poisson Correction | Example 3: Uniform Baseline | Returned Object | Other Families | References

Terminology | Example | Key Arguments | criterion | sampling.method | likelihood | boot and B | Outputs | Returned object | References

Model and Rare-Event Setting | How ssp.relogit() Works | Pilot Estimator | Choosing criterion | criterion = "optA" | criterion = "optL" | criterion = "LCC" | criterion = "uniform" | Negative-Sampling Probabilities | Choosing likelihood | likelihood = "weighted" | likelihood = "logOddsCorrection" | $$P(Y_i = 1 \mid x_i, i \in S) | Method Summary | Example Data | Example 1: Default Rare-Event Fit | Example 2: A-Optimal Criterion | Example 3: Uniform Baseline | Returned Object | References

Terminology | Example | Key Arguments | criterion | sampling.method | likelihood | constraint | Outputs | References

Full-Data Problem | Step 1: Pilot Estimation | Step 2: Calculate Raw Importance Scores | Step 3: Construct Probabilities and Draw the Second-Step Subsample | Step 4: Fit a Corrected Subsample Objective Function | Step 5: Optional Combination | $$\widehat | Function Map | Checking Points