--- title: "2. Mathematical Details" output: bookdown::html_document2: base_format: rmarkdown::html_vignette fig_caption: yes toc: true toc_depth: 2 number_sections: true pkgdown: as_is: true vignette: > %\VignetteIndexEntry{2. Mathematical Details} %\VignetteEngine{knitr::rmarkdown} %\VignetteEncoding{UTF-8} bibliography: references.bib --- ```{r, include = FALSE} knitr::opts_chunk$set( collapse = TRUE, comment = "#>" ) ``` # Introduction Specification document - a mathematical description of models used by bage. Note: some features described here have not been implemented yet. # Input data {#sec:data} - outcome variable: events, numbers of people, or some sort of measure on a continuous variable such as income or expenditure - exposure/size/weights - disagg by one or more variables. Almost always includes age, sex/gender, and time. May include other variables eg region, ethnicity, education. - not all combinations of variables present; may be some missing values # Models ## Poisson likelihood {#sec:pois} Let $y_i$ be a count of events in cell $i = 1, \cdots, n$ and let $w_i$ be the corresponding exposure measure, with the possibility that $w_i \equiv 1$. The likelihood under the Poisson model is then \begin{align} y_i & \sim \text{Poisson}(\gamma_i w_i) (\#eq:lik-pois-1) \\ \gamma_i & \sim \text{Gamma}\left(\xi^{-1}, (\mu_i \xi)^{-1}\right), (\#eq:lik-pois-2) \end{align} using the shape-rates parameterisation of the Gamma distribution. Parameter $\xi$ governs dispersion, with \begin{equation} \text{var}(\gamma_i \mid \mu_i, \xi) = \xi \mu_i^2 \end{equation} and \begin{equation} \text{var}(y_i \mid \mu_i, \xi, w_i) = (1 + \xi \mu_i w_i ) \times \mu_i w_i. \end{equation} We allow $\xi$ to equal 0, in which case the model reduces to \begin{equation} y_i \sim \text{Poisson}(\mu_i w_i). \end{equation} For $\xi > 0$, Equations \@ref(eq:lik-pois-1) and \@ref(eq:lik-pois-2) are equivalent to \begin{equation} y_i \sim \text{NegBinom}\left(\xi^{-1}, (1 + \mu_i w_i \xi)^{-1}\right) \end{equation} [@norton2018sampling; @simpson2022priors]. This is the format we use internally for estimation. When values for $\gamma_i$ are needed, we generate them on the fly, using the fact that \begin{equation} \gamma_i \mid y_i, w_i, \mu_i, \xi \sim \text{Gamma}\left(y_i + \xi^{-1}, w_i + (\xi \mu_i)^{-1}\right). \end{equation} ## Binomial likelihood {#sec:binom} The likelihood under the binomial model is \begin{align} y_i & \sim \text{Binomial}(w_i, \gamma_i) (\#eq:lik-binom-1) \\ \gamma_i & \sim \text{Beta}\left(\xi^{-1} \mu_i, \xi^{-1}(1 - \mu_i)\right). (\#eq:lik-binom-2) \end{align} Parameter $\xi$ again governs dispersion, with \begin{equation} \text{var}(\gamma_i \mid \mu_i, \xi) = \frac{\xi}{1 + \xi} \times \mu_i (1 -\mu_i) \end{equation} and \begin{equation} \text{var}(y_i \mid w_i, \mu_i, \xi) = \frac{\xi w_i + 1}{\xi + 1} \times w_i \mu_i (1 - \mu_i). \end{equation} We allow $\xi$ to equal 0, in which case the model reduces to \begin{equation} y_i \sim \text{Binom}(w_i, \mu_i). \end{equation} Equations \@ref(eq:lik-binom-1) and \@ref(eq:lik-binom-2) are equivalent to \begin{equation} y_i \sim \text{BetaBinom}\left(w_i, \xi^{-1} \mu_i, \xi^{-1} (1 - \mu_i) \right), \end{equation} which is what we use internally. Values for $\gamma_i$ can be generated using \begin{equation} \gamma_i \mid y_i, w_i, \mu_i, \xi \sim \text{Beta}\left(y_i + \xi^{-1} \mu_i, w_i - y_i + \xi^{-1}(1-\mu_i) \right). \end{equation} ## Normal likelihood {#sec:norm} The normal model is \begin{align} y_i & \sim \text{N}(\gamma_i, w_i^{-1} \sigma^2) \\ \gamma_i & = \bar{y} + s \mu_i \\ \sigma^2 & = \bar{w} s^2 \xi^2 \end{align} where \begin{align} \bar{y} & = \frac{\sum_{i=1}^n y_i}{n} \\ s & = \sqrt{\frac{\sum_{i=1}^n (y_i - \bar{y})^2}{n-1}} \\ \bar{w} & = \frac{\sum_{i=1}^n w_i}{n}. \end{align} Mean $\mu_i$ and stanard deviation $\xi$ are defined on scale that used standardized versinos of $y$ and $w$. Standardizing allows us to apply the same priors as we use for the Poisson and binomial models. # Model for prior means {#sec:means} Let $\pmb{\mu} = (\mu_1, \cdots, \mu_n)^{\top}$. Our model for $\pmb{\mu}$ is \begin{equation} \pmb{\mu} = \sum_{m=0}^{M} \pmb{X}^{(m)} \pmb{\beta}^{(m)} + \pmb{Z} \pmb{\zeta} (\#eq:means) \end{equation} where - $\beta^{(0)}$ is an intercept; - $\pmb{\beta}^{(m)}$, $m=1,\cdots,M$ is a vector with $J_m$ elements describing a main effect or interaction formed from the dimensions of data $\pmb{y}$; - $\pmb{X}^{(m)}$ is an $n \times J_m$ matrix of 1s and 0s, the $i$th row of which picks out the element of $\pmb{\beta}^{(m)}$ that is used with cell $i$; - $\pmb{Z}$ is a $n \times P$ matrix of covariates; and - $\pmb{\zeta}$ is a coefficient vector with $P$ elements. # Priors for Intercept, Main Effects, and Interactions {#sec:priors} ## General features ### 'Along' and 'by' dimensions Each $\pmb{\beta}^{(m)}$, $m > 0$, can be a main effect, involving a single dimension, or an interaction, involving two dimensions. Some priors, when applied to an interaction, treat one dimension, referred to as the 'along' dimension, differently from the remaining dimensions, referred to as 'by' dimensions. A random walk prior (Section \@ref(sec:pr-rw)), for instance, consists of an independent random walk along the 'along' dimension, within each combination of the 'by' dimensions. We use $v = 1, \cdots, V_m$ to denote position within the 'along' dimension, and $u = 1, \cdots, V_m$ to denote position within a classification formed by the 'by' dimensions. When there are no sum-to-zero constraints (see below), $U_m = \prod_k d_k$ where $d_k$ is the number of elements in the $k$th 'by' variable. When there are sum-to-zero constraints, $U_m = \prod_k (d_k - 1)$. If a prior involves an 'along' dimension but the user does not specify one, the procedure for choosing a dimension is as follows: - if the term involves time, use the time dimension; - otherwise, if the term involves age, use the age dimension; - otherwise, raise an error asking the user to explicitly specify a dimension. ### Constraints {#sec:constraints} With some combinations of terms and priors, some $\pmb{\beta}^{(m)}$ are only weakly identified, and have diffuse prior distributions. Even when this happens, however, the quantity $\mu_i = \sum_{m=0}^M \beta_{j_i^m}^{(m)}$ is still well identified, so the weak identification may not matter to the aims of the analysis. If, however, stronger identification is required, it can be achieved by imposing constraints on the elements of the $\pmb{\beta}^{(m)}$. This is done via the `con` argument. At present only two choices for `con` have been implemented. The first is `"none"`, where no constraints are applied. This is the default. The second is `"by"`. The `"by"` option can only be used if $\pmb{\beta}^{(m)}$ has an 'along' dimension. If `con` is `"by"`, then within each element $v$ of the 'along' dimension, the sum of the $\beta_j^{(m)}$ across each 'by' dimension is zero. For instance, if $\pmb{\beta}^{(m)}$ is an interaction between time, region, and sex, with time as the 'along' variable, then within each combination of time and region, the values for females and males sum to zero, and within each combination of time and sex, the values for regions sum to zero. Except in the case of dynamic SVD-based priors (eg Sections \@ref(sec:pr-svd-rw)), `"by"` constraints are implemented internally by drawing values within an unrestricted lower-dimensional space, and then transforming to the restricted higher-dimensional space. For instance, a random walk prior for a time-region interaction with $R$ regions consists of $R-1$ unrestricted random walks along time, which are converted into $R$ random walks that sum to zero across region. Matrices for transforming between the unrestricted and restricted spaces are constructed using the QR decomposition, as described in Section 1.8.1 of @wood2017generalized. With dynamic SVD-based priors, we draw values for the SVD coefficients with no constraints, convert these to unconstrained values for $\pmb{\beta}^{(m)}$, and then subtract means. ### Algorithm for assigning default priors - If $\pmb{\beta}^{(m)}$ has one or two elements, assign $\pmb{\beta}^{(m)}$ a fixed-normal prior (Section \@ref(sec:pr-fnorm)); - otherwise, if $\pmb{\beta}^{(m)}$ involves time, assign $\pmb{\beta}^{(m)}$ a random walk prior (Section \@ref(sec:pr-rw)) along the time dimension; - otherwise, if $\pmb{\beta}^{(m)}$ involves age, assign $\pmb{\beta}^{(m)}$ a random walk prior (Section \@ref(sec:pr-rw)) along the age dimension; - otherwise, assign $\pmb{\beta}^{(m)}$ a normal prior (Section \@ref(sec:pr-norm)) The intercept term $\pmb{\beta}^{(0)}$ can only be given a fixed-normal prior (Section \@ref(sec:pr-fnorm)) or a Known prior (Section \@ref(sec:pr-known)). ## N() {#sec:pr-norm} ### Model Exchangeable normal \begin{align} \beta_j^{(m)} & \sim \text{N}\left(0, \tau_m^2 \right) \\ \tau_m & \sim \text{N}^+\left(0, A_{\tau}^{(m)2}\right) \end{align} ### Contribution to posterior density \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \prod_{j=1}^{J_m} \text{N}(\beta_j^{(m)} \mid 0, \tau_m^2) \end{equation} ### Forecasting \begin{equation} \beta_{J_m+h+1}^{(m)} \sim \text{N}(0, \tau_m^2) \end{equation} ### Code ``` N(s = 1) ``` - `s` is $A_{\tau}^{(m)}$. Defaults to 1. ## NFix() {#sec:pr-fnorm} ### Model Exchangeable normal, with fixed standard deviation \begin{equation} \beta_j^{(m)} \sim \text{N}\left(0, A_{\beta}^{(m)2}\right) \end{equation} ### Contribution to posterior density \begin{equation} \prod_{j=1}^{J_m} \text{N}(\beta_j^{(m)} \mid 0, A_{\beta}^{(m)2}) \end{equation} ### Forecasting \begin{equation} \beta_{J_m+h+1}^{(m)} \sim \text{N}(0, A_{\beta}^{(m)2}) \end{equation} ### Code ``` NFix(sd = 1) ``` - `sd` is $A_{\tau}^{(m)}$. Defaults to 1. ## RW() {#sec:pr-rw} ### Model Random walk \begin{align} \beta_{u,1}^{(m)} & \sim \text{N}\left(0, (A_0^{(m)})^2\right) \\ \beta_{u,v}^{(m)} & \sim \text{N}(\beta_{u,v-1}^{(m)}, \tau_m^2), \quad v = 2, \cdots, V_m \\ \tau_m & \sim \text{N}^+\left(0, (A_{\tau}^{(m)})^2\right) \end{align} $A_0^{(m)}$ can be 0, implying that $\beta_{u,1}^{(m)}$ is fixed at 0. When $U_m > 1$, constraints (Section \@ref(sec:constraints)) can be applied. ### Contribution to posterior density \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \prod_{u=1}^{U_m} \text{N}\left(\beta_{u,1}^{(m)} \mid 0, (A_0^{(m)})^2\right) \prod_{v=2}^{V_m} \text{N}\left(\beta_{u,v}^{(m)} \mid \beta_{u,v-1}^{(m)}, \tau_m^2 \right) \end{equation} ### Forecasting \begin{equation} \beta_{u,V_m+h}^{(m)} \sim \text{N}(\beta_{u,V_m+h-1}^{(m)}, \tau_m^2) \end{equation} If the prior includes sum-to-zero constraints, means are subtracted from the forecasted values within each combination of 'along' and 'by' variables. ### Code ``` RW(s = 1, along = NULL, con = c("none", "by")) ``` - `s` is $A_{\tau}^{(m)}$. Defaults to 1. - `sd` is $A_0^{(m)}$. Defaults to 1. - `along` used to identify 'along' and 'by' dimensions. - if `con` is `"by"`, sum-to-zero constraints are applied. ## RW2() {#sec:pr-rw2} ### Model Second-order random walk \begin{align} \beta_{u,1}^{(m)} & \sim \text{N}\left(0, (A_0^{(m)})^2\right) \\ \beta_{u,2}^{(m)} & \sim \text{N}\left(\beta_{u,1}, (A_{\eta}^{(m)})^2\right) \\ \beta_{u,v}^{(m)} & \sim \text{N}\left(2 \beta_{u,v-1}^{(m)} - \beta_{u,v-2}^{(m)}, \tau_m^2\right), \quad v = 3, \cdots, V_m \\ \tau_m & \sim \text{N}^+\left(0, (A_{\tau}^{(m)})^2\right) \end{align} $A_0^{(m)}$ can be 0, implying that $\beta_{u,1}^{(m)}$ is fixed at 0. When $U_m > 1$, constraints (Section \@ref(sec:constraints)) can be applied. ### Contribution to posterior density \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \prod_{u=1}^{U_m} \text{N}(\beta_{u,1}^{(m)} \mid 0, (A_0^{(m)})^2) \text{N}(\beta_{u,2}^{(m)} \mid \beta_{u,1}^{(m)}, (A_{\eta}^{(m)})^2) \prod_{v=3}^{V_m} \text{N}\left(\beta_{u,v}^{(m)} \mid 2 \beta_{u,v-1}^{(m)} - \beta_{u,v-2}^{(m)}, \tau_m^2 \right) \end{equation} ### Forecasting \begin{equation} \beta_{u,V_m+h}^{(m)} \sim \text{N}(2 \beta_{u,V_m+h-1}^{(m)} - \beta_{u,V_m+h-2}^{(m)}, \tau_m^2) \end{equation} If the prior includes sum-to-zero constraints, means are subtracted from the forecasted values within each combination of 'along' and 'by' variables. ### Code ``` RW2(s = 1, sd = 1, sd_slope = 1, along = NULL, con = c("none", "by")) ``` - `s` is $A_{\tau}^{(m)}$ - `sd` is $A_0^{(m)}$ - `sd_slope` is $A_{\eta}^{(m)}$ - `along` used to identify 'along' and 'by' dimensions - if `con` is `"by"`, sum-to-zero constraints are applied ## RW2_Infant() {#sec:pr-rw2-infant} ### Model Second-order random walk with infant indicator. Designed for age profiles for mortality rates. Along dimension must be age. \begin{align} \beta_{u,1}^{(m)} & \sim \text{N}(0, 1) \\ \beta_{u,2}^{(m)} & \sim \text{N}\left(0, (A_{\eta}^{(m)})^2\right) \\ \beta_{u,3}^{(m)} & \sim \text{N}\left(2 \beta_{u,2}^{(m)}, \tau_m^2\right) \\ \beta_{u,v}^{(m)} & \sim \text{N}\left(2 \beta_{u,v-1}^{(m)} - \beta_{u,v-2}^{(m)}, \tau_m^2\right), \quad v = 4, \cdots, V_m \\ \tau_m & \sim \text{N}^+\left(0, (A_{\tau}^{(m)})^2\right) \end{align} When $U_m > 1$, constraints (Section \@ref(sec:constraints)) can be applied. ### Contribution to posterior density \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \prod_{u=1}^{U_m} \text{N}(\beta_{u,1}^{(m)} \mid 0, 1) \text{N}(\beta_{u,2}^{(m)} \mid 0, (A_{\eta}^{(m)})^2) \text{N}\left(\beta_{u,3}^{(m)} \mid 2 \beta_{u,2}^{(m)}, \tau_m^2 \right) \prod_{v=4}^{V_m} \text{N}\left(\beta_{u,v}^{(m)} \mid 2 \beta_{u,v-1}^{(m)} - \beta_{u,v-2}^{(m)}, \tau_m^2 \right) \end{equation} ### Forecasting Terms with an `RW2_Infant()` prior cannot be forecasted. ### Code ``` RW2_Infant(s = 1, sd_slope = 1, con = c("none", "by")) ``` - `s` is $A_{\tau}^{(m)}$ - `sd_slope` is $A_{\eta}^{(m)}$ - if `con` is `"by"`, sum-to-zero constraints are applied ## RW_Seas() {#sec:rwseas} ### Model Random walk with seasonal effect \begin{align} \beta_{u,v}^{(m)} & = \alpha_{u,v} + \lambda_{u,v}, \quad v = 1, \cdots, V_m \\ \alpha_{u,1}^{(m)} & \sim \text{N}\left(0, (A_0^{(m)})^2 \right) \\ \alpha_{u,v}^{(m)} & \sim \text{N}(\alpha_{u,v-1}^{(m)}, \tau_m^2), \quad v = 2, \cdots, V_m \\ \lambda_{u,v}^{(m)} & \sim \text{N}(0, A_{\lambda}^{(m)}), \quad v = 1, \cdots, S_m - 1 \\ \lambda_{u,v}^{(m)} & = -\sum_{s=1}^{S_m-1} \lambda_{u,v-s}^{(m)}, \quad v = S_m,\; 2S_m, \cdots \\ \lambda_{u,v}^{(m)} & \sim \text{N}(\lambda_{u,v-S_m}^{(m)}, \omega_m^2), \quad \text{otherwise} \\ \tau_m & \sim \text{N}^+\left(0, A_{\tau}^{(m)2}\right) \\ \omega_m & \sim \text{N}^+\left(0, A_{\omega}^{(m)2}\right) \end{align} $A_0^{(m)}$ can be 0, implying that $\alpha_{u,1}^{(m)}$ is fixed at 0. $A_{\omega}^{(m)2}$ can be set to zero, implying that seasonal effects are fixed over time. When $U_m > 1$, constraints (Section \@ref(sec:constraints)) can be applied. ### Contribution to posterior density \begin{align} & \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \text{N}(\omega_m \mid 0, A_{\omega}^{(m)2}) \notag \\ & \quad \times \prod_{u=1}^{U_m} \bigg( \text{N}\left(\alpha_{u,1}^{(m)} \mid 0, (A_0^{(m)})^2 \right) \prod_{v=2}^{V_m} \text{N}\left(\alpha_{u,v}^{(m)} \mid \alpha_{u,v-1}^{(m)}, \tau_m^2 \right) \notag \\ & \quad \times \prod_{v=1}^{S_m-1} \text{N}\left(\lambda_{u,v}^{(m)} \mid 0, (A_{\lambda}^{(m)})^2\right) \prod_{\substack{v > S_m \\ (v-1) \bmod S_m \neq 0}}^{V_m} \text{N}\left(\lambda_{u,v}^{(m)} \mid \lambda_{u,v-S_m}^{(m)}, \omega_m^2\right) \bigg) \end{align} ### Forecasting \begin{align} \alpha_{J_m+h}^{(m)} & \sim \text{N}(\alpha_{J_m+h-1}^{(m)}, \tau_m^2) \\ \lambda_{J_m+h}^{(m)} & \sim \text{N}(\lambda_{J_m+h-S_m}^{(m)}, \omega_m^2) \\ \beta_{J_m+h}^{(m)} & = \alpha_{J_m+h}^{(m)} + \lambda_{J_m+h}^{(m)} \end{align} ### Code ``` RW_Seas(n_seas, s = 1, sd = 1, s_seas = 0, sd_seas = 1, along = NULL, con = c("none", "by")) ``` - `n_seas` is $S_m$ - `s` is $A_{\tau}^{(m)}$ - `sd` is $A_0^{(m)}$ - `s_seas` is $A_{\omega}^{(m)}$ - `sd_seas` is $A_{\lambda}^{(m)}$ - `along` used to identify 'along' and 'by' dimensions - if `con` is `"by"`, sum-to-zero constraints are applied ## RW2_Seas() {#sec:rw2seas} ### Model Second-order random work, with seasonal effect \begin{align} \beta_{u,v}^{(m)} & = \alpha_{u,v} + \lambda_{u,v}, \quad v = 1, \cdots, V_m \\ \alpha_{u,1}^{(m)} & \sim \text{N}\left(0, (A_0^{(m)})^2 \right) \\ \alpha_{u,2}^{(m)} & \sim \text{N}\left(\alpha_{u,1}^{(m)}, (A_{\eta}^{(m)})^2\right) \\ \alpha_{u,v}^{(m)} & \sim \text{N}(2 \alpha_{u,v-1}^{(m)} - \alpha_{u,v-2}^{(m)}, \tau_m^2), \quad v = 3, \cdots, V_m \\ \lambda_{u,v}^{(m)} & \sim \text{N}(0, A_{\lambda}^{(m)}), \quad v = 1, \cdots, S_m - 1 \\ \lambda_{u,v}^{(m)} & = -\sum_{s=1}^{S_m-1} \lambda_{u,v}^{(m)}, \quad v = S_m,\; 2S_m, \cdots \\ \lambda_{u,v}^{(m)} & \sim \text{N}(\lambda_{u,v-S_m}^{(m)}, \omega_m^2), \quad \text{otherwise} \\ \tau_m & \sim \text{N}^+\left(0, A_{\tau}^{(m)2}\right) \\ \omega_m & \sim \text{N}^+\left(0, A_{\omega}^{(m)2}\right) \end{align} $A_0^{(m)}$ can be 0, implying that $\alpha_{u,1}^{(m)}$ is fixed at 0. $A_{\omega}^{(m)2}$ can be set to zero, implying that seasonal effects are fixed over time. When $U_m > 1$, constraints (Section \@ref(sec:constraints)) can be applied. ### Contribution to posterior density \begin{align} & \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \text{N}(\omega_m \mid 0, A_{\omega}^{(m)2}) \notag \\ & \times \prod_{u=1}^{U_m} \bigg( \text{N}(\alpha_{u,1}^{(m)} \mid 0, (A_0^{(m)})^2 ) \text{N}(\alpha_{u,2}^{(m)} \mid \alpha_{u,1}^{(m)}, (A_{\eta}^{(m)})^2 ) \notag \\ & \quad \times \prod_{v=3}^{V_m} \text{N}(\alpha_{u,v}^{(m)} \mid 2 \alpha_{u,v-1}^{(m)} - \alpha_{u,v-2}^{(m)}, \tau_m^2 ) \notag \\ & \quad \times \prod_{v=1}^{S_m-1} \text{N}(\lambda_{u,v}^{(m)} \mid 0, (A_{\lambda}^{(m)})^2) \notag \\ & \quad \times \prod_{\substack{v > S_m \\ (v-1) \bmod S_m \neq 0}}^{V_m} \text{N}(\lambda_{u,v}^{(m)} \mid \lambda_{u,v-S_m}^{(m)}, \omega_m^2) \bigg) \end{align} ### Forecasting \begin{align} \alpha_{J_m+h}^{(m)} & \sim \text{N}(2 \alpha_{J_m+h-1}^{(m)} - \alpha_{J_m+h-2}^{(m)}, \tau_m^2) \\ \lambda_{J_m+h}^{(m)} & \sim \text{N}(\lambda_{J_m+h-S_m}^{(m)}, \omega_m^2) \\ \beta_{J_m+h}^{(m)} & = \alpha_{J_m+h}^{(m)} + \lambda_{J_m+h}^{(m)} \end{align} ### Code ``` RW2_Seas(n_seas, s = 1, sd = 1, sd_slope = 1, s_seas = 0, sd_seas = 1, along = NULL, con = c("none", "by")) ``` - `n_seas` is $S_m$ - `s` is $A_{\tau}^{(m)}$ - `sd` is $A_0^{(m)}$ - `sd_slope` is $A_{\eta}^{(m)}$ - `s_seas` is $A_{\omega}^{(m)}$ - `sd_seas` is $A_{\lambda}^{(m)}$ - `along` used to identify 'along' and 'by' dimensions - if `con` is `"by"`, sum-to-zero constraints are applied ## AR() {#sec:pr-ar} ### Model \begin{equation} \beta_{u,v}^{(m)} \sim \text{N}\left(\phi_1^{(m)} \beta_{u,v-1}^{(m)} + \cdots + \phi_{K_m}^{(m)} \beta_{u,v-{K_m}}^{(m)}, \omega_m^2\right), \quad v = K_m + 1, \cdots, V_m. \end{equation} Internally, TMB derives values for $\beta_{u,v}^{(m)}, v = 1, \cdots, K_m$, and for $\omega_m$, that imply a stationary distribution, and that give every term $\beta_{u,v}^{(m)}$ the same marginal variance. We denote this marginal variance $\tau_m^2$, and assign it a prior \begin{equation} \tau_m \sim \text{N}^+(0, A_{\tau}^{(m)2}). \end{equation} Each of the $\phi_k^{(m)}$ has prior \begin{equation} \frac{\phi_k^{(m)} + 1}{2} \sim \text{Beta}(S_1^{(m)}, S_2^{(m)}). \end{equation} ### Contribution to posterior density \begin{equation} \text{N}^+\left(\tau_m \mid 0, A_{\tau}^{(m)2} \right) \prod_{k=1}^{K_m} \text{Beta}\left(\tfrac{1}{2} \phi_k^{(m)} + \tfrac{1}{2} \mid 2, 2 \right) \prod_{u=1}^{U_m} p\left( \beta_{u,1}^{(m)}, \cdots, \beta_{u,V_m}^{(m)} \mid \phi_1^{(m)}, \cdots, \phi_{K_m}^{(m)}, \tau_m \right) \end{equation} where $p\left( \beta_{u,1}^{(m)}, \cdots, \beta_{u,V_m}^{(m)} \mid \phi_1^{(m)}, \cdots, \phi_{K_m}^{(m)}, \tau_m \right)$ is calculated internally by TMB. ### Forecasting \begin{equation} \beta_{u,V_m + h}^{(m)} \sim \text{N}\left(\phi_1^{(m)} \beta_{u,V_m + h - 1}^{(m)} + \cdots + \phi_{K_m}^{(m)} \beta_{u,V_m+h-K_m}^{(m)}, \tau_m^2\right) \end{equation} ### Code ``` AR(n_coef = 2, s = 1, shape1 = 5, shape2 = 5, along = NULL, con = c("none", "by")) ``` - `n_coef` is $K_m$ - `s` is $A_{\tau}^{(m)}$ - `shape1` is $S_1^{(m)}$ - `shape2` is $S_2^{(m)}$ - `along` is used to indentify the 'along' and 'by' dimensions ## AR1() {#sec:pr-ar1} Special case or `AR()`, with extra options for autocorrelation coefficient. ### Model \begin{align} \beta_{u,1}^{(m)} & \sim \text{N}(0, \tau_m^2) \\ \beta_{u,v}^{(m)} & \sim \text{N}(\phi_m \beta_{u,v-1}^{(m)}, (1 - \phi_m^2) \tau_m^2) \quad v = 2, \cdots, V_m \\ \phi_m & = a_{0,m} + (a_{1,m} - a_{0,m}) \phi_m^{\prime} \\ \phi_m^{\prime} & \sim \text{Beta}(S_1^{(m)}, S_2^{(m)}) \\ \tau_m & \sim \text{N}^+\left(0, A_{\tau}^{(m)2}\right). \end{align} This is adapted from the specification used for AR1 densities in [TMB](http://kaskr.github.io/adcomp/classdensity_1_1AR1__t.html). It implies that the marginal variance of all $\beta_{u,v}^{(m)}$ is $\tau_m^2$. We require that $-1 < a_{0m} < a_{1m} < 1$. ### Contribution to posterior density \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \text{Beta}( \phi_m^{\prime} \mid S_1^{(m)}, S_2^{(m)}) \prod_{u=1}^{U_m} \text{N}\left(\beta_{u,1}^{(m)} \mid 0, \tau_m^2 \right) \prod_{u=1}^{U_m} \prod_{j=2}^{V_m} \text{N}\left(\beta_{u,v}^{(m)} \mid \phi_m \beta_{u,v-1}^{(m)}, (1 - \phi_m^2) \tau_m^2 \right) \end{equation} ### Forecasting \begin{equation} \beta_{J_m + h}^{(m)} \sim \text{N}\left(\phi_m \beta_{J_m + h - 1}^{(m)}, (1 - \phi_m^2) \tau_m^2\right) \end{equation} ### Code ``` AR1(s = 1, shape1 = 5, shape2 = 5, min = 0.8, max = 0.98, along = NULL, con = c("none", "by")) ``` - `s` is $A_{\tau}^{(m)}$ - `shape1` is $S_1^{(m)}$ - `shape2` is $S_2^{(m)}$ - `min` is $a_{0m}$ - `max` is $a_{1m}$ - `along` is used to identify 'along' and 'by' dimensions The defaults for `min` and `max` are based on the defaults for function `ets()` in R package **forecast** [@hyndman2008automatic]. ## Lin() {#sec:pr-lin} ### Model \begin{align} \beta_{u,v}^{(m)} & = \alpha_{u,v}^{(m)} + \epsilon_{u,v}^{(m)} \\ \alpha_{u,v}^{(m)} & = \left(v - \frac{V_m + 1}{2}\right) \eta_u^{(m)} \\ \eta_u^{(m)} & \sim \text{N}\left(B_{\eta}^{(m)}, (A_{\eta}^{(m)})^2\right)\\ \epsilon_{u,v}^{(m)} & \sim \text{N}(0, \tau_m^2) \\ \tau_m & \sim \text{N}^+\left(0, (A_{\tau}^{(m)})^2\right) \end{align} Note that $\sum_{v=1}^{V_m} \alpha_{u,v}^{(m)} = 0$. ### Contribution to posterior density \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \prod_{u=1}^{U_m} \text{N}(\eta_u^{(m)} \mid B_{\eta}^{(m)}, A_{\eta}^{(m)2}) \prod_{u=1}^{U_m} \prod_{v=1}^{V_m} \text{N}\left(\beta_{u,v}^{(m)} \mid v - \frac{V_m + 1}{2}, \tau_m^2 \right) \end{equation} ### Forecasting \begin{equation} \beta_{u,V_m + h}^{(m)} \sim \text{N}\left(\left(\frac{V_m - 1}{2}+ h\right) \eta_u^{(m)}, \tau_m^2\right) \end{equation} ### Code ``` Lin(s = 1, mean_slope = 0, sd_slope = 1, along = NULL, con = c("none", "by")) ``` - `s` is $A_{\tau}^{(m)}$ - `mean_slope` is $B_{\eta}^{(m)}$ - `sd_slope` is $A_{\eta}^{(m)}$ - `along` is used to indentify 'along' and 'by' dimensions - if `con` is `"by"`, sum-to-zero constraints are applied ## Lin_AR() {#sec:pr-lin-ar} ### Model \begin{align} \beta_{u,v}^{(m)} & = \alpha_{u,v}^{(m)} + \epsilon_{u,v}^{(m)} \\ \alpha_{u,v}^{(m)} & = \left(v - \frac{V_m + 1}{2}\right) \eta_u^{(m)} \\ \eta_u^{(m)} & \sim \text{N}\left(B_{\eta}^{(m)}, (A_{\eta}^{(m)})^2\right)\\ \epsilon_{u,v}^{(m)} & \sim \text{N}\left(\phi_1^{(m)} \epsilon_{u,v-1}^{(m)} + \cdots + \phi_{K_m}^{(m)} \epsilon_{u,v-{K_m}}^{(m)}, \omega_m^2\right), \quad v = K_m + 1, \cdots, V_m \end{align} Note that $\sum_{v=1}^{V_m} \alpha_{u,v}^{(m)} = 0$. Internally, TMB derives values for $\epsilon_{u,v}^{(m)}, v = 1, \cdots, K_m$, and for $\omega_m$, that provide the $\epsilon_{u,v}^{(m)}$ with a stationary distribution in which each term has the same marginal variance. We denote this marginal variance $\tau_m^2$, and assign it a prior \begin{equation} \tau_m \sim \text{N}^+(0, A_{\tau}^{(m)2}). \end{equation} Each of the individual $\phi_k^{(m)}$ has prior \begin{equation} \frac{\phi_k^{(m)} + 1}{2} \sim \text{Beta}(S_1^{(m)}, S_2^{(m)}). \end{equation} ### Contribution to posterior density \begin{align} & \text{N}^+\left(\tau_m \mid 0, A_{\tau}^{(m)2} \right) \prod_{k=1}^{K_m} \text{Beta}\left( \tfrac{1}{2} \phi_k^{(m)} + \tfrac{1}{2} \mid S_1^{(m)}, S_2^{(m)} \right) \notag \\ & \quad \times \prod_{u=1}^{U_m} \text{N}(\eta_u^{(m)} \mid 0, A_{\eta}^{(m)2}) p\left( \epsilon_{u,1}^{(m)}, \cdots, \epsilon_{u,V_m}^{(m)} \mid \phi_1^{(m)}, \cdots, \phi_{K_m}^{(m)}, \tau_m \right) \end{align} where $p\left( \epsilon_{u,1}^{(m)}, \cdots, \epsilon_{u,V_m}^{(m)} \mid \phi_1^{(m)}, \cdots, \phi_{K_m}^{(m)}, \tau_m \right)$ is calculated internally by TMB. ### Forecasting \begin{align} \beta_{u, V_m + h}^{(m)} & = \left(\frac{V_m - 1}{2}+ h\right) \eta_u^{(m)} + \epsilon_{u,V_m+h}^{(m)} \\ \epsilon_{u,V_m+h}^{(m)} & \sim \text{N}\left(\phi_1^{(m)} \epsilon_{u,V_m + h - 1}^{(m)} + \cdots + \phi_{K_m}^{(m)} \epsilon_{u,V_m+h-K_m}^{(m)}, \omega_m^2\right) \end{align} ### Code ``` Lin_AR(n_coef = 2, s = 1, shape1 = 5, shape2 = 5, mean_slope = 0, sd_slope = 1, along = NULL, con = c("none", "by")) ``` - `n_coef` is $K_m$ - `s` is $A_{\tau}^{(m)}$ - `shape1` is $S_1^{(m)}$ - `shape2` is $S_2^{(m)}$ - `mean_slope` is $B_{\eta}^{(m)}$ - `sd_slope` is $A_{\eta}^{(m)}$ - `along` is used to indentify 'along' and 'by' variables - if `con` is `"by"`, sum-to-zero constraints are applied ## Lin_AR1() {#sec:pr-lin-ar1} ### Model \begin{align} \beta_{u,v}^{(m)} & = \alpha_{u,v}^{(m)} + \epsilon_{u,v}^{(m)} \\ \alpha_{u,v}^{(m)} & = \left(v - \frac{V_m + 1}{2}\right) \eta_u^{(m)} \\ \eta_u^{(m)} & \sim \text{N}\left(B_{\eta}^{(m)}, (A_{\eta}^{(m)})^2\right)\\ \epsilon_{u,1}^{(m)} & \sim \text{N}\left(0, \tau_m^2 \right) \\ \epsilon_{u,v}^{(m)} & \sim \text{N}\left(\phi_m \epsilon_{u,v-1}^{(m)}, (1 - \phi_m^2) \tau_m^2 \right), \quad v = 2, \cdots, V_m \\ \phi_m & = a_{0,m} + (a_{1,m} - a_{0,m}) \phi_m^{\prime} \\ \phi_m^{\prime} & \sim \text{Beta}(S_1^{(m)}, S_2^{(m)}) \\ \tau_m & \sim \text{N}^+\left(0, A_{\tau}^{(m)2}\right). \end{align} Note that $\sum_{v=1}^{V_m} \alpha_{u,v}^{(m)} = 0$. ### Contribution to posterior density \begin{align} & \text{N}^+\left(\tau_m \mid 0, A_{\tau}^{(m)2} \right) \text{Beta}( \phi_m^{\prime} \mid S_1^{(m)}, S_2^{(m)}) \notag \\ & \quad \times \prod_{u=1}^{U_m} \text{N}(\eta_u^{(m)} \mid 0, A_{\eta}^{(m)2}) \text{N}\left(\epsilon_{u,1}^{(m)} \mid 0, \tau_m^2 \right) \prod_{v=2}^{V_m} \text{N}\left(\epsilon_{u,v}^{(m)} \mid \phi_m \epsilon_{u,v-1}^{(m)}, (1 - \phi_m^2) \tau_m^2 \right) \end{align} ### Forecasting \begin{align} \beta_{u, V_m + h}^{(m)} & = \left(\frac{V_m - 1}{2}+ h\right) \eta_u^{(m)} + \epsilon_{u,V_m+h}^{(m)} \\ \epsilon_{u,V_m+h}^{(m)} & \sim \text{N}\left(\phi_m \epsilon_{u,V_m + h - 1}^{(m)}, (1 - \phi_m^2) \tau_m^2\right) \end{align} ### Code ``` Lin_AR1(s = 1, shape1 = 5, shape2 = 5, min = 0.8, max = 0.98, mean_slope = 0, sd_slope = 1, along = NULL, con = c("none", "by")) ``` - `s` is $A_{\tau}^{(m)}$ - `shape1` is $S_1^{(m)}$ - `shape2` is $S_2^{(m)}$ - `min` is $a_{0m}$ - `max` is $a_{1m}$ - `mean_slope` is $B_{\eta}^{(m)}$ - `sd_slope` is $A_{\eta}^{(m)}$ - `along` is used to indentify 'along' and 'by' variables - if `con` is `"by"`, sum-to-zero constraints are applied ## Sp() {#sec:pr-spline} ### Model Penalised spline (P-spline) \begin{equation} \pmb{\beta}_u^{(m)} = \pmb{B}^{(m)} \pmb{\alpha}_u^{(m)}, \quad u = 1, \cdots, U_m \end{equation} where $\pmb{\beta}_u^{(m)}$ is the subvector of $\pmb{\beta}^{(m)}$ composed of elements from the $u$th combination of the 'by' variables, $\pmb{B}^{(m)}$ is a $V_m \times K_m$ matrix of B-splines, and $\pmb{\alpha}_u^{(m)}$ has a second-order random walk prior (Section \@ref(sec:pr-rw2)). $\pmb{B}^{(m)} = (\pmb{b}_1^{(m)}(\pmb{v}), \cdots, \pmb{b}_{K_m}^{(m)}(\pmb{v}))$, with $\pmb{v} = (1, \cdots, V_m)^{\top}$. The B-splines are centered, so that $\pmb{1}^{\top} \pmb{b}_k^{(m)}(\pmb{v}) = 0$, $k = 1, \cdots, K_m$. ### Contribution to posterior density \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \prod_{u=1}^{U_m} \prod_{k=1}^2 \text{N}(\alpha_{u,k}^{(m)} \mid 0, 1) \prod_{u=1}^{U_m}\prod_{k=3}^{K_m} \text{N}\left(\alpha_{u,k}^{(m)} - 2 \alpha_{u,k-1}^{(m)} + \alpha_{u,k-2}^{(m)} \mid 0, \tau_m^2 \right) \end{equation} ### Forecasting Terms with a `Sp()` prior cannot be forecasted. ### Code ``` Sp(n = NULL, s = 1) ``` - `n` is $K_m$. Defaults to $\max(0.7 J_m, 4)$. - `s` is the $A_{\tau}^{(m)}$ from the second-order random walk prior. Defaults to 1. - `along` is used to identify 'along' and 'by' variables ## SVD() {#sec:pr-svd} ### Model **Age but no sex or gender** Let $\pmb{\beta}_u$ be the age effect for the $u$th combination of the 'by' variables. With an SVD prior, \begin{equation} \pmb{\beta}_u^{(m)} = \pmb{F}^{(m)} \pmb{\alpha}_u^{(m)} + \pmb{g}^{(m)}, \quad u = 1, \cdots, U_m \end{equation} where $\pmb{F}^{(m)}$ is a $V_m \times K_m$ matrix, and $\pmb{g}^{(m)}$ is a vector with $V_m$ elements, both derived from a singular value decomposition (SVD) of an external dataset of age-specific values for all sexes/genders combined. The construction of $\pmb{F}^{(m)}$ and $\pmb{g}^{(m)}$ is described in Appendix \@ref(app:svd). The centering and scaling used in the construction allow use of the simple prior \begin{equation} \alpha_{u,k}^{(m)} \sim \text{N}(0, 1), \quad u = 1, \cdots, U_m, k = 1, \cdots, K_m. \end{equation} **Joint model of age and sex/gender** In the joint model, vector $\pmb{\beta}_u$ represents the interaction between age and sex/gender for the $u$th combination of the 'by' variables. Matrix $\pmb{F}^{(m)}$ and vector $\pmb{g}^{(m)}$ are calculated from data that separate sexes/genders. The model is otherwise unchanged. **Independent models for each sex/gender** In the independent model, vector $\pmb{\beta}_{s,u}$ represents age effects for sex/gender $s$ and the $u$th combination of the 'by' variables, and we have \begin{equation} \pmb{\beta}_{s,u}^{(m)} = \pmb{F}_s^{(m)} \pmb{\alpha}_{s,u}^{(m)} + \pmb{g}_s^{(m)}, \quad s = 1, \cdots, S; \quad u = 1, \cdots, U_m \end{equation} Matrix $\pmb{F}_s^{(m)}$ and vector $\pmb{g}_s^{(m)}$ are calculated from data that separate sexes/genders. The prior is \begin{equation} \alpha_{s,u,k}^{(m)} \sim \text{N}(0, 1), \quad s = 1, \cdots, S; \quad u = 1, \cdots, U_m; \quad k = 1, \cdots, K_m. \end{equation} ### Contribution to posterior density \begin{equation} \prod_{u=1}^{U_m}\prod_{k=1}^{K_m} \text{N}\left(\alpha_{uk}^{(m)} \mid 0, 1 \right) \end{equation} for the age-only and joint models, and \begin{equation} \prod_{s=1}^S \prod_{u=1}^{U_m}\prod_{k=1}^{K_m} \text{N}\left(\alpha_{s,u,k}^{(m)} \mid 0, 1 \right) \end{equation} for the independent model ### Forecasting Terms with an SVD prior cannot be forecasted. ### Code ``` SVD(ssvd, n_comp = NULL, indep = TRUE) ``` where - `ssvd` is an object containing $\pmb{F}$ and $\pmb{g}$ - `n_comp` is the number of components to be used (which defaults to `ceiling(n/2)`, where `n` is the number of components in `ssvd` - `indep` determines whether and independent or joint model will be used if the term being modelled contains a sex or gender variable. ## SVD_RW() {#sec:pr-svd-rw} ### Model The `SVD_RW()` prior is identical to the `SVD()` prior except that the coefficients evolve over time, following independent random walks. For instance, in the combined-sex/gender and joint models with $K_m$ SVD components, \begin{align} \pmb{\beta}_{u,t}^{(m)} & = \pmb{F}^{(m)} \pmb{\alpha}_{u,t}^{(m)} + \pmb{g}^{(m)} \\ \alpha_{u,k,1}^{(m)} & \sim \text{N}(0, (A_0^{(m)})^2), \\ \alpha_{u,k,t}^{(m)} & \sim \text{N}(\alpha_{u,k,t-1}^{(m)}, \tau_m^2), \quad t = 2, \cdots, T \\ \tau_m & \sim \text{N}^+\left(0, (A_{\tau}^{(m)})^2\right) \end{align} ### Contribution to posterior density In the combined-sex/gender and joint models, \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \prod_{u=1}^{U_m} \prod_{k=1}^{K_m} \text{N}(\alpha_{u,k,1}^{(m)} \mid 0, (A_0^{(m)})^2) \prod_{t=2}^{T} \text{N}\left(\alpha_{u,k,t}^{(m)} \mid \alpha_{u,k,t-1}^{(m)}, \tau_m^2 \right), \end{equation} and in the independent model, \begin{equation} \text{N}(\tau_m \mid 0, A_{\tau}^{(m)2}) \prod_{u=1}^{U_m} \prod_{s=1}^{S} \prod_{k=1}^{K_m} \text{N}(\alpha_{u,s,k,1}^{(m)} \mid 0, (A_0^{(m)})^2) \prod_{t=2}^{T} \text{N}\left(\alpha_{u,s,k,t}^{(m)} \mid \alpha_{u,s,k,t-1}^{(m)}, \tau_m^2 \right) \end{equation} ### Forecasting \begin{align} \alpha_{u,k,T+h}^{(m)} & \sim \text{N}(\alpha_{u,k,T+h-1}^{(m)}, \tau_m^2) \\ \pmb{\beta}_{u,T+h}^{(m)} & = \pmb{F}^{(m)} \pmb{\alpha}_{u,T+h}^{(m)} + \pmb{g}^{(m)} \end{align} ### Code ``` SVD_RW(ssvd, n_comp = NULL, indep = TRUE, s = 1, sd = 1, con = c("none", "by")) ``` where - `ssvd` is an object containing $\pmb{F}$ and $\pmb{g}$ - `n_comp` is $K_m$ - `indep` determines whether and independent or joint model will be used if the term being modelled contains a sex or gender variable. - `s` is $A_{\tau}^{(m)}$ - `sd` is $A_0^{(m)}$ ## SVD_RW2(), SVD_AR(), SVD_AR1() {#sec:pr-svd-oth} The `SVD_RW2()`, `SVD_AR()` and `SVD_AR1()` priors have the same structure as the `SVD_RW()` prior, but with `RW2()`, `AR()`, and `AR1()` priors for the along dimension taking the place of the `RW()` prior. ## Known {#sec:pr-known} ### Model Elements of $\pmb{\beta}^{(m)}$ are treated as known with certainty. ### Contribution to posterior density Known priors make no contribution to the posterior density. ### Forecasting Main effects with a known prior cannot be forecasted. ### Code ``` Known(values) ``` - `values` is a vector containing the $\beta_j^{(m)}$. # Covariates {#sec:pr-cov} ## Model Matrix $\pmb{Z}$ is a standardized version of the original covariate data supplied by the user. We standardize every numeric variable to have mean 0 and standard deviation one. We then convert categorical variables to sets of indicator variables using R's 'treatment' contrast. If a variable in the original data is categorical with $C$ categories, then it is converted into $C-1$ indicator variables, with the first category as the omitted variable. The elements of $\pmb{\zeta}$ have prior \begin{equation} \zeta_p \sim \text{N}(0, 1) \end{equation} ## Contribution to posterior density \begin{equation} \prod_{p=1}^P \text{N}(\zeta_p | 0, 1) \end{equation} ## Forecasting A model with covariates can be used for forecasting provided that - the coefficients (the $\zeta_p$) are non-time-varying, - future values for the covariates (the columns of $\pmb{Z}$) can be inferred from the classifying variables (other than time), or are supplied by the user. ## Code ``` set_covariates(mod, formula) ``` - `mod` Object of class `"bage_mod"` - `formula` One-sided R formula describing the covariates to be used # Prior for dispersion terms ## Model Use exponential distribution, parameterised using mean, \begin{equation} \xi \sim \text{Exp}(\mu_{\xi}) \end{equation} ## Contribution to prior density \begin{equation} p(\xi) = \frac{1}{\mu_{\xi}} \exp\left(\frac{-\xi}{\mu_{\xi}}\right) \end{equation} ## Code ``` set_disp(mean = 1) ``` - `mean` is $\mu_{\xi}$ # Data models ## Data models for outcome ### Random Rounding to Base 3 Random rounding to base 3 (RR3) is a confidentialization method used by some statistical agencies. It is applied to counts data. Each count $x$ is rounded randomly as follows: - If $x \mod 3 = 0$, then $x$ is left unchanged; - if $x \mod 3 = 1$ then $x$ is changed to $x-1$ with probability 2/3, and is changed to $x + 2$ with probability 1/3; and - if $x \mod 3 = 2$ then $x$ is changed to $x-2$ with probability 1/3, and is changed to $x + 1$ with probability 2/3. RR3 data models can be used with Poisson or binomial likelihoods. Let $y_i$ denote the observed value for the outcome, and $y_i^*$ the true value. The likelihood with a RR3 data model is then \begin{align} p(y_i | \gamma_i, w_i) & = \sum_{y_i^*} p(y_i | y_i^*) p(y_i^* | \gamma_i, w_i) \\ & = \sum_{k = -2, -1, 0, 1, 2} p(y_i | y_i + k) p(y_i + k | \gamma_i, w_i) \\ & = \tfrac{1}{3} p(y_i - 2 | \gamma_i, w_i) + \tfrac{2}{3} p(y_i - 1 | \gamma_i, w_i) + p(y_i | \gamma_i, w_i) + \tfrac{2}{3} p(y_i + 1 | \gamma_i, w_i) + \tfrac{1}{3} p(y_i + 2 | \gamma_i, w_i) \end{align} ## Data models for exposure or size # Estimation ## Filtering and Aggregation {#sec:filter-ag} The data that we supply to TMB is a a filtered and aggregated version of the data that the user provides through the `data` argument. In the filtering stage, we remove any rows where (i) the offset is 0 or NA, or (ii) the outcome variable is NA. In the aggregation stage, we identify any rows in the data that duplicated combinations of classification variables. For instance, if the classification variables are `age` and `sex`, and we have two rows where `age` is `"20-24"` and sex is `"Female"`, then these rows would count as duplicated combinations. We aggregate offset and outcome variables across these duplicates. With Poisson and binomial models, the aggregation formula for outcomes is \begin{equation} y^{\text{new}} = \sum_{i=1}^D y_i^{\text{old}}, \end{equation} and the aggregation formula for exposure/size is \begin{equation} w^{\text{new}} = \sum_{i=1}^D w_i^{\text{old}}, \end{equation} where $D$ is the number of times a particular combination is duplicated. With normal models, the aggregation formula for outcomes is \begin{equation} y^{\text{new}} = \frac{\sum_{i=1}^D y_i^{\text{old}}}{D}, \end{equation} and the aggregation formuala for weights is \begin{equation} w^{\text{new}} = \frac{1}{\sum_{i=1}^D \frac{1}{w_i^{\text{old}}}}. \end{equation} ## Inner-Outer Approximation ### Step 0: Select 'inner' and 'outer' variables Select variables to be used in inner model. By default, these are the age, sex, and time variables in the model. All remaining variables are 'outer' variables. ### Step 1: Fit inner model Aggregate the data using the classification formed by the inner variables. (See Section \@ref(sec:filter-ag) on aggregation procedures.) Remove all terms not involving 'inner' variables, other than the intercept term, from the model. Set dispersion to 0. Fit the resulting model. ### Step 2: Fit outer model Let $\hat{\mu}_i^{\text{in}}$ be point estimates for the linear predictor $\mu_i$ obtained from the inner model. **Poisson model** Aggregate the data using the classification formed by the outer variables. Remove all terms involving the 'inner' variables, plus the intercept, from the model. Set dispersion to 0. Set exposure to $w_i^{\text{out}} = \hat{\mu}^{\text{in}} w_i$. Fit the model. **Binomial model** Fit the original model, but set dispersion to 0, and for all terms from the 'inner' model, use Known priors using point estimates from the inner model. **Normal model** Aggregate the data using the classification formed by the outer variables. Remove all terms involving the 'inner' variables, plus the intercept, from the model. Set the outcome variable to to $y_i^{\text{out}} = y_i - \hat{\mu}^{\text{in}}. Fit the model. ### Step 4: Concatenate estimates Concatenate posterior distributions for the inner terms from the inner model to posterior distributions for the outer terms from the outer model. ### Step 5: Calculate dispersion If the original model includes a dispersion term, then estimate dispersion. Let $\hat{\mu}_i^{\text{comb}}$ be point estimates for the linear predictor obtained from the concatenated estimates. **Poisson model** Use the original disaggregated data, or, if the original data contains more then 10,000 rows, select 10,000 rows at random from the original data. Remove all terms from the original model except for the intercept. Set exposure to $w_i^{\text{out}} = \hat{\mu}^{\text{comb}} w_i$. **Binomial model** Fit the the original model, but with all terms except the intercept having Known priors, where the values are obtained from point estimates from the concatenated estimates. **Normal model** Use the original disaggregated data, or, if the original data contains more then 10,000 rows, select 10,000 rows at random from the original data. Remove all terms from the original model except for the intercept. Set the outcome to $y_i^{\text{out}} = y_i - \hat{\mu}^{\text{comb}}$. # Deriving outputs Running TMB yields a set of means $\pmb{m}$, and a precision matrix $\pmb{Q}^{-1}$, which together define the approximate joint posterior distribution of - intercept, main effects, and internactions $\pmb{\beta}^{(m)}$, $m = 0, \cdots, M$, - typically, hyper-parameters for the $\pmb{\beta}^{(m)}$, in many cases transformed to another scale, such as a log scale, - optionally, dispersion term $\xi$, and - optionally, covariate cofficients vector $\pmb{zeta}$. Let $\tilde{\pmb{\theta}}$ to denote a vector containing all these quantities. We draw values for $\tilde{\pmb{\theta}} \mid \pmb{m}, \pmb{Q}^{-1}$, typically using sparse matrix methods (as implemented in package **sparseMVN**. Next we convert any transformed hyper-parameters back to the original units, and insert values for $\pmb{\beta}^{(m)}$ for terms that have Known priors. We denote the resulting vector $\pmb{\theta}$. We draw from the distribution of $\pmb{\gamma} \mid \pmb{y}, \pmb{\theta}$ using the methods described in Sections \@ref(sec:pois)-\@ref(sec:norm). Output from the Normal model receives special treatment. As described in section \@ref(sec:norm) and \@ref(sec:means), the Normal model is \begin{align} \frac{y_i - \bar{y}}{s} & \sim \text{N}\left(\mu_i, \frac{\bar{w}}{w_i} \xi^2\right) \\ \mu_i & = \sum_{m=0}^{M} \beta_{j_i^m}^{(m)} + (\pmb{Z} \pmb{\zeta})_i \end{align} # Simulation To generate one set of simulated values, we start with values for exposure, trials, or weights, $\pmb{w}$, and possibly covariates $\pmb{Z}$, then go through the following steps: 1. Draw values for any parameters in the priors for the $\pmb{\beta}^{(m)}$, $m = 1, \cdots, M$. 1. Conditional on the values drawn in Step 1, draw values the $\pmb{\beta}^{(m)}$, $m = 0, \cdots, M$. 1. If the model contains seasonal effects, draw the standard deviation $\kappa_m$, and then the effects $\pmb{\lambda}^{(m)}$. 1. If the model contains covariates, draw $\varphi$ and $\vartheta_p$ where necessary, draw coefficient vector $\pmb{\zeta}$. 1. Use values from steps 2--4 to form the linear predictor $\sum_{m=0}^{M} \pmb{X}^{(m)} (\pmb{\beta}^{(m)} + \pmb{\lambda}^{(m)}) + \pmb{Z} \pmb{\zeta}$. 1. Back-transform the linear predictor, to obtain vector of cell-specific parameters $\pmb{\mu}$. 1. If the model contains a dispersion parameter $\xi$, draw values from the prior for $\xi$. 1. In Poisson and binomial models, use $\pmb{\mu}$ and, if present, $\xi$ to draw $\pmb{\gamma}$. 1. In Poisson and binomial models, use $\pmb{\gamma}$ and $\pmb{w}$ to draw $\pmb{y}$; in normal models, use $\pmb{\mu}$, $\xi$, and $\pmb{w}$ to draw $\pmb{y}$. # Replicate data ## Model ### Poisson likelihood #### Condition on $\pmb{\gamma}$ \begin{equation} y_i^{\text{rep}} \sim \text{Poisson}(\gamma_i w_i) \end{equation} #### Condition on $(\pmb{\mu}, \xi)$ \begin{align} y_i^{\text{rep}} & \sim \text{Poisson}(\gamma_i^{\text{rep}} w_i) \\ \gamma_i^{\text{rep}} & \sim \text{Gamma}(\xi^{-1}, (\xi \mu_i)^{-1}) \end{align} which is equivalent to \begin{equation} y_i^{\text{rep}} \sim \text{NegBinom}\left(\xi^{-1}, (1 + \mu_i w_i \xi)^{-1}\right) \end{equation} ### Binomial likelihood #### Condition on $\pmb{\gamma}$ \begin{equation} y_i^{\text{rep}} \sim \text{Binomial}(w_i, \gamma_i) \end{equation} #### Condition on $(\pmb{\mu}, \xi)$ \begin{align} y_i^{\text{rep}} & \sim \text{Binomial}(w_im \gamma_i^{\text{rep}}) \\ \gamma_i^{\text{rep}} & \sim \text{Beta}\left(\xi^{-1} \mu_i, \xi^{-1}(1 - \mu_i)\right) \end{align} ### Normal likelihood \begin{equation} y_i^{\text{rep}} \sim \text{N}(\gamma_i, \xi^2 / w_i) \end{equation} ### Data models for outcomes If the overall model includes a data model for the outcome, then a further set of draws is made, deriving values for the observed outcomes, given values for the true outcomes. ## Code ``` replicate_data(x, condition_on = c("fitted", "expected"), n = 20) ``` # Appendices ## Definitions {#app:defn} | Quantity | Definition | |:---------|:----------------------------------------------------| | $i$ | Index for cell, $i = 1, \cdots, n$. | $y_i$ | Value for outcome variable. | | $w_i$ | Exposure, number of trials, or weight. | | $\gamma_i$ | Super-population rate, probability, or mean. | | $\mu_i$ | Cell-specific mean. | | $\xi$ | Dispersion parameter. | | $g()$ | Log, logit, or identity function. | | $m$ | Index for intercept, main effect, or interaction. $m = 0, \cdots, M$. | | $j$ | Index for element of a main effect or interaction. | | $u$ | Index for combination of 'by' variables for an interaction. $u = 1, \cdots U_m$. $U_m V_m = J_m$ | | $v$ | Index for the 'along' dimension of an interaction. $v = 1, \cdots V_m$. $U_m V_m = J_m$ | | $\beta^{(0)}$ | Intercept. | | $\pmb{\beta}^{(m)}$ | Main effect or interaction. $m = 1, \cdots, M$. | | $\beta_j^{(m)}$ | $j$th element of $\pmb{\beta}^{(m)}$. $j = 1, \cdots, J_m$. | | $\pmb{X}^{(m)}$ | Matrix mapping $\pmb{\beta}^{(m)}$ to $\pmb{y}$. | | $\pmb{Z}$ | Matrix of covariates. | | $\pmb{\zeta}$ | Parameter vector for covariates $\pmb{Z}^{(m)}$. | | $A_0$ | Scale parameter in prior for intercept $\beta^{(0)}$ or initial value. | | $\tau_m$ | Standard deviation parameter for main effect or interaction. | | $A_{\tau}^{(m)}$ | Scale parameter in prior for $\tau_m$. | | $\pmb{\alpha}^{(m)}$ | Parameter vector for P-spline and SVD priors. | | $\alpha_k^{(m)}$ | $k$th element of $\pmb{\alpha}^{(m)}$. $k = 1, \cdots, K_m$. | | $\pmb{V}^{(m)}$ | Covariance matrix for multivariate normal prior. | | $h_j^{(m)}$ | Linear covariate | | $\eta^{(m)}$ | Parameter specific to main effect or interaction $\pmb{\beta}^{(m)}$. | | $\eta_u^{(m)}$ | Parameter specific to $u$th combination of 'by' variables in interaction $\pmb{\beta}^{(m)}$. | | $A_{\eta}^{(m)}$ | Standard deviation in normal prior for $\eta_m$. | | $\omega_m$ | Standard deviation of parameter $\eta_c$ in multivariate priors. | | $\phi_m$ | Correlation coefficient in AR1 densities. | | $a_{0m}$, $a_{1m}$ | Minimum and maximum values for $\phi_m$. | | $\pmb{B}^{(m)}$ | B-spline matrix in P-spline prior. | | $\pmb{b}_k^{(m)}$ | B-spline. $k = 1, \cdots, K_m$. | | $\pmb{F}^{(m)}$ | Matrix in SVD prior. | | $\pmb{g}^{(m)}$ | Offset in SVD prior. | | $\pmb{\beta}_{\text{trend}}$ | Trend effect. | | $\pmb{\beta}_{\text{cyc}}$ | Cyclical effect. | | $\pmb{\beta}_{\text{seas}}$ | Seasonal effect. | | $\varphi$ | Global shrinkage parameter in shrinkage prior. | | $A_{\varphi}$ | Scale term in prior for $\varphi$. | | $\vartheta_p$ | Local shrinkage parameter in shrinkage prior. | | $p_0$ | Expected number of non-zero coefficients in $\pmb{\zeta}$. | | $\hat{\sigma}$ | Empirical scale estimate in prior for $\varphi$. | | $\pi$ | Vector of hyper-parameters | ## SVD prior for age {#app:svd} Let $\pmb{A}$ be a matrix of age-specific estimates from an international database, transformed to take values in the range $(-\infty, \infty)$. Each column of $\pmb{A}$ represents one set of age-specific estimates, such as log mortality rates in Japan in 2010, or logit labour participation rates in Germany in 1980. Let $\pmb{U}$, $\pmb{D}$, $\pmb{V}$ be the matrices from a singular value decomposition of $\pmb{A}$, where we have retained the first $K$ components. Then \begin{equation} \pmb{A} \approx \pmb{U} \pmb{D} \pmb{V}. (\#eq:svd1) \end{equation} Let $m_k$ and $s_k$ be the mean and sample standard deviation of the elements of the $k$th row of $\pmb{V}$, with $\pmb{m} = (m_1, \cdots, m_k)^{\top}$ and $\pmb{s} = (s_1, \cdots, s_k)^{\top}$. Then \begin{equation} \tilde{\pmb{V}} = (\text{diag}(\pmb{s}))^{-1} (\pmb{V} - \pmb{m} \pmb{1}^{\top}) \end{equation} is a standardized version of $\pmb{V}$. We can rewrite \@ref(eq:svd1) as \begin{align} \pmb{A} & \approx \pmb{U} \pmb{D} (\text{diag}(\pmb{s}) \tilde{\pmb{V}} + \pmb{m} \pmb{1}^{\top}) \\ & = \pmb{F} \tilde{\pmb{V}} + \pmb{g} \pmb{1}^{\top}, (\#eq:svd2) \end{align} where $\pmb{F} = \pmb{U} \pmb{D} \text{diag}(\pmb{s})$ and $\pmb{g} = \pmb{U} \pmb{D} \pmb{m}$. Let $\tilde{\pmb{v}}_l$ be a randomly-selected column from $\tilde{\pmb{V}}$. From the construction of $\tilde{\pmb{V}}$ we have $\text{E}[\tilde{v}_{kl}] = 0$ and $\text{var}[\tilde{v}_{kl}] = 1$. If $\pmb{z}$ is a vector of standard normal variables, then \begin{equation} \pmb{F} \pmb{z} + \pmb{g} \end{equation} should look approximately like a randomly-selected column from the original data matrix $\pmb{A}$. # References