No articles match
mfp2 User Guide7 days ago
Introduction to the mfp2 package | Supported model families | Choosing an interface | Estimation algorithm | Installation | Linear regression | Fitting MFP models using the formula interface | A short matrix-interface example | Shifting and scaling of covariates | Manually specifying shifts and scales | Centering and prediction on the original predictor scale | Subsetting and analysis populations | Factors and grouped terms after subsetting | Setting degrees of freedom for each variable | Tuning parameters for MFP | Nominal significance levels | Information criteria | Retaining predictors and forcing functional forms | Model comparison tests | Choosing FP powers using subject-matter knowledge | Specifying candidate powers | Candidate powers and selected powers | Explanation of output from the model-selection algorithm | Methods defined for mfp2 | Inspecting the selected model | Summarising the fitted effects | Extracting and interpreting coefficients | Making predictions | Graphical presentation of FP functions | Fitted functions and residuals | Contrasts relative to a reference value | Choosing where to evaluate the function | Handling categorical variables | Unordered factors | Ordered factors | Formula-interface example | Categorical predictors in the matrix interface | Logistic regression | Survival data | Fitting a Cox model | Using a monotonic function for positive lymph nodes | Stratified Cox model | Examples for additional outcome families | Multinomial regression | Ordinal regression | Generalized estimating equations | References
ACD Transformations with mfp217 days ago
Overview | Definition of the ACD transformation | Inverse normal transformation of ranks | Power-linear approximation | Back-transformation to the interval (0, 1) | Interpretation | ACD in the fractional polynomial family | Function-selection procedure for ACD (FSPA) | P-value selection | Information criteria | Simulation example | Fitting an ACD model | Matrix interface | Formula interface | Comparison with standard MFP | Plotting the fitted functions | Practical remarks | References
MFP Methodology and Practical Considerations17 days ago
MFP methodology and practical considerations | Overview of MFP | Fractional polynomial models for a continuous variable | Function selection procedure (FSP) | Multivariable MFP procedure | Backfitting cycles | Key issues and approaches to handling them | The variable has to be positive | Sample size and influential observations | Lack of fit | Local features | Model-selection uncertainty and limitations | References
Modelling Spike-at-Zero Covariates with Fractional Polynomials17 days ago
Introduction | Prostate cancer data | Univariable analysis with p-values | Forcing inclusion or the full SAZ model | Multivariable analysis | Selection with AIC or BIC | Simulated data: continuous + binary | Several SAZ covariates in one model | Matrix interface | Further details | Paper reference
Multivariable Fractional Polynomial Interaction Models17 days ago
Introduction | Advanced prostate cancer data | Interaction without adjustment | Plotting the interaction | Predictions at chosen ages | Interaction with adjustment for prognostic factors | Selection with AIC | Several interaction candidates | Fixed linear adjustment model | More than two groups | The xy interface | Further details | Paper reference