Variabelegewijze imputatiefouten

In de vorige oefening heb je de geschatte imputatiefouten uit de output van missForest gehaald. Dat leverde twee getallen op:

de genormaliseerde wortel van de gemiddelde kwadratische fout (NRMSE) voor alle continue variabelen;
het aandeel fout geclassificeerde waarden (PFC) voor alle categorische variabelen.

Het kan echter best dat het imputatiemodel voor de ene continue variabele uitstekend presteert en voor een andere juist slecht! Om zulke gevallen te onderzoeken, kun je missForest variabelegewijze foutschattingen laten produceren. Dat doe je door het argument variablewise op TRUE te zetten.

De biopics-gegevens en de missForest-package zijn al voor je geladen, dus laten we de fouten eens van dichterbij bekijken!

Deze oefening maakt deel uit van de cursus

Omgaan met missende data met imputaties in R

Cursus bekijken

Praktische interactieve oefening

Probeer deze oefening eens door deze voorbeeldcode in te vullen.

# Impute biopics data with missForest computing per-variable errors
imp_res <- ___(___, ___ = ___)

Code bewerken en uitvoeren

Deze oefening maakt deel uit van de cursus

Omgaan met missende data met imputaties in R

SkillTag.level.advancedSkillTag.label

4.7+

Begin de cursus gratis

In this chapter, you’ll find out why missing data can be a risk when analyzing a dataset. You’ll be introduced to the three missing data mechanisms and learn how to recognize them using statistical tests and visualization tools.

Exercise 1: Missing data: what can go wrong Exercise 2: Linear regression with incomplete data Exercise 3: Analyzing regression output Exercise 4: Comparing models Exercise 5: Missing data mechanisms Exercise 6: Recognizing missing data mechanisms Exercise 7: t-test for MAR: data preparation Exercise 8: t-test for MAR: interpretation Exercise 9: Visualizing missing data patterns Exercise 10: Aggregation plot Exercise 11: Spine plot Exercise 12: Mosaic plot

Get to know the taxonomy of imputation methods and learn three donor-based techniques: mean, hot-deck, and k-Nearest-Neighbors imputation. You’ll look under the hood to see how these methods work, before learning how to apply them to a real-world tropical weather dataset. Along the way, you’ll also learn useful tricks that you can use to make them work even better for your problems.

Exercise 1: Mean imputation Exercise 2: Smelling the danger of mean imputation Exercise 3: Mean-imputing the temperature Exercise 4: Assessing imputation quality with margin plot Exercise 5: Hot-deck imputation Exercise 6: Vanilla hot-deck Exercise 7: Hot-deck tricks & tips I: imputing within domains Exercise 8: Hot-deck tricks & tips II: sorting by correlated variables Exercise 9: k-Nearest-Neighbors imputation Exercise 10: Choosing the number of neighbors Exercise 11: kNN tricks & tips I: weighting donors Exercise 12: kNN tricks & tips II: sorting variables

It’s time to learn how to use statistical and machine learning models, such as linear regression, logistic regression, and random forests, to impute missing data. In this chapter, you’ll look into how the models make their predictions and use this knowledge to draw the imputed values from conditional distributions. This is important as it ensures your imputations are more varied and plausible, making them more similar to the true data.

Exercise 1: Modelgebaseerde imputatiemethode Exercise 2: Imputatie met lineaire regressie Exercise 3: Missende waarden initialiseren en over variabelen itereren Exercise 4: Convergentie detecteren Exercise 5: Variatie in data repliceren Exercise 6: Imputatie met logistic regression Exercise 7: Trekken uit een conditionele verdeling Exercise 8: Modelgebaseerde imputatie met meerdere variabeletype Exercise 9: Boomgebaseerde imputatie Exercise 10: Imputeren met random forests Exercise 11: Variabelegewijze imputatiefouten

Huidige oefening

Exercise 12: Afweging tussen snelheid en nauwkeurigheid

Imputed values are not set in stone. They are just estimates and estimates come with some uncertainty. In this final chapter, you’ll discover how bootstrapping and chained equation using the mice package can be used to incorporate imputation uncertainty into your models and analyses to make them more reliable and robust.

Exercise 1: Multiple imputation by bootstrapping Exercise 2: Wrapping imputation & modeling in a function Exercise 3: Running the bootstrap Exercise 4: Bootstrapping confidence intervals Exercise 5: Multiple imputation by chained equations Exercise 6: The mice flow: mice - with - pool Exercise 7: Choosing default models Exercise 8: Using predictor matrix Exercise 9: Putting it all together Exercise 10: Analyzing missing data patterns Exercise 11: Imputing and inspecting outcomes Exercise 12: Inference with imputed data Exercise 13: Final remarks