Use an information criterion to select a local log-linear model

1 2 |

`densi` |
A matrix with one row and 2^k-1 column containing cell counts or empirical cell probabilities corresponding to all the possible capture patterns. |

`models` |
A list of character vectors, with each vector containing
column names from the associated log-linear design matrix.
For example, see the output of |

`N` |
If you multiply |

`ic` |
The information criterion, such as AIC, AICc, BIC, or BICpi. |

`averaging` |
Logical: TRUE means that we use information criterion scores to do model averaging. |

`normalized` |
Logical: TRUE means that beta0 will be adjusted so that the log-linear model corresponds to cell probabilities instead of expected cell counts. |

`rasch` |
Logical: TRUE means that the Rasch model is a candidate. |

Just like `flat.IC`

except that it is designed to take in a local
average instead of a full capture-recapture dataset

`pred` |
Estimated rate of missingness for the selected model |

`form` |
Formula of the selected model |

Zach Kurtz

Questions? Problems? Suggestions? Tweet to @rdrrHQ or email at ian@mutexlabs.com.

All documentation is copyright its authors; we didn't write any of that.