Bayes (Hierarcical Bayesian model): Difference between revisions

From Opasnet
Jump to navigation Jump to search
mNo edit summary
 
(19 intermediate revisions by 3 users not shown)
Line 1: Line 1:
{{method}}
This page describes utilization of Hierarcical Bayesin method. The following page briefly introduces theory behind the method, and illustrates the use the method by examples.  
This page describes utilization of Hierarcical Bayesin method. The following page briefly introduces theory behind the method, and illustrates the use the method by examples.  
<br>
= Introduction<br>  =


== The Bayesian method  ==
== The Bayesian method  ==


The Bayes method is based on the bayesian rule. The following equations can be found from the reference 1, pages 7-8. The Bayesian rule is:<br>
The Bayes method is based on the bayesian rule. The following equations can be found from the reference 1, pages 7-8. The Bayesian rule is:
 
<br>


p(θ,y) = p(θ)p(y|θ)  
p(θ,y) = p(θ)p(y|θ)  
<br>


,where p(θ) is prior distribution and on p(y|θ) is distribution acquired from the data. By contidionalizing θ with y (which is known), we get:  
,where p(θ) is prior distribution and on p(y|θ) is distribution acquired from the data. By contidionalizing θ with y (which is known), we get:  
<br>


p(θ|y) = p(θ)p(y|θ) / p(y)  
p(θ|y) = p(θ)p(y|θ) / p(y)  


<br>
By conditionalizing this way, we obtain p(?|y), which is prior probability conditionalised by the data. This is also called posterior probability. With these equations, we can adjust presuppositions about the probability density function of the data using data itself. As an end result we obtain upgraded probability density function, which is called posterior density.
 
By conditionalizing this way, we obtain p(θ|y), which is prior probability conditionalised by the data. This is also called posterior probability. With these equations, we can adjust presuppositions about probability density function of the data using data itself. As a end result we obtain upgraded probability density function, which is called posterior density.&nbsp;&nbsp;&nbsp;
 
<br>
 
Posterior distribution ~ prior distribution x distribution estimated by the data<br>
 
<br>


== <br>Hierarcical Bayesian method  ==
Posterior distribution ~ prior distribution x distribution estimated by the data


Hierarkisen bayesin idea on helpoin ymmärtää konkreettisen esimerkin kautta. Meidän tapauksessa havaitaan pitoisuuksia eri alueittain, joiden voidaan olettaa olevan alueittain samankaltaisia. Oletetaan, että pitoisuudet alueittain ovat otoksia alueiden yhteisestä pitoisuus jakaumasta. Näin ollen voimme käyttää havaittuja pitoisuuksia yli alueiden estimoitaessa tietyn alueen pitoisuuksia. Tämä tapahtuu ehdollistamalla havaittavissa olevia arvoja tiettyihin parametreihin. Näille parametreille oletetaan todennäköisyydet uusien parametrien avulla, joita kutsutaan hyperparametreiksi.
===Hierarcical Bayesian method  ===


Kirjassa Bayesian Data Analysis second edition sivuilla 117-121 löytyy kattava selvennys ideasta ja sivulla 119 on kuva hierarkisesta mallista.  
We create an example of the method in this article. We study dioxin concentrations in different sea area which we assume to be share a similar form of distribution. In other words, we assume that the concentraions in different areas are samples of the total concentration distribution. Thus, we can use observations of concentrations in all areas to estimate concentrations in one particular area of interest. This means conditionalizing observated values to certain parameters. We assume probabilites to these parameters with new parameters, which are called hyperparameters.  


Käyttötarkoitus meillä on, että pyritään löytämään keinoja mallintaa tilastollisin keinoin pitoisuuksia kaloista. Tavoitteena saada aikaan malli, jonka avulla pienestäkin datasta saataisiin suhteellisen pitävä arvio koko alueen kalapopulaatiolle.  
There is an extensive explanation of the idea and the method in reference 1, pages 117-121, and you can find a picture of the method in page 119. In our example, we try to create a model which gives us a relatively accurate concentration estimate for all different areas.  


<br>
== Material and methods  ==


<br>
=== Methods  ===


= Material and methods  =
The method used herein is called hierarcical Bayes method. Theory can be read from reference 1. We used WinBUGS and R as a modelling tool.
 
== Methods  ==
 
Käytettävä menetelmä on hierarkkinen bayes. Teoria kirjasta Bayesian data analysis second edition. Työkaluna WinBUGS ja R.  


http://www.mrc-bsu.cam.ac.uk/bugs/  
http://www.mrc-bsu.cam.ac.uk/bugs/  


Tästä osoitteesta voi hakea päivityksen ja lisenssin. Päivitysohjeet ja lisenssiohjeet kattavat, eli tässä ei pitäisi tulla ongelmia. Tekstin vain kopioi WinBUGS uuteen tiedostoon ja tools valikosta decode kuten ohjeissa mainitaan .  
You can get the program and updates from the address above. They also include quite extensive manuals. You need to copy the text to WinBUGS to a new file, and select decode from tools menu and follow the manual.&nbsp;


Minkä nimisiä ovat muuttujat mallissa, taulukko (muuttujan nimi ja parametrit)<br>Mallin graafinen kuva (vaikka sama kuva kuin WinBUGS:ssa) muuttujien vuorovaikutussuhteista<br><br>
Table 1. Names of the variables and parameters in the model.&nbsp; 


Table 1. Names of the variables and parameters in the model.&nbsp;<br>
{| {{prettytable}}
 
{| width="200" cellspacing="1" cellpadding="1" border="1"
|-
|-
| muuttuja<br>
| Variable 
| jakauma
| Distribution
| tyyppi<br>
| Nature 
| parametrit
| Parameters
| arvo
| Definition
| Kuvaus<br>
| Description
|-
|-
| alpha.tau  
| alpha.tau  
| gamma  
| gamma  
| stokastinen
| stocastic
| 10^-3  
| 10^-3  
| <br>
|  
| Tarkkuusparametri
| Parameter of accuracy
|-
|-
| alpha.c  
| alpha.c  
| norm  
| norm  
| stokastinen
| stocastic
| (0,10^-6)  
| (0,10^-6)  
| <br>
|  
| Keskiarvo
| Mean
|-
|-
| alpha0  
| alpha0  
| <br>
|  
| laskettu
| calculated
| <br>
|  
| (a.c)-(b.c)*xbar  
| (a.c)-(b.c)*xbar  
| Lähtötaso
| Starting level
|-
|-
| alpha[i]  
| alpha[i]  
| norm  
| norm  
| stokastinen
| stocastic
| (a.c,a.t)  
| (a.c,a.t)  
| <br>
|  
| Vakiotaso
| Constant term
|-
|-
| beta.tau  
| beta.tau  
| gamma  
| gamma  
| stokastinen
| stocastic
| 10^-3  
| 10^-3  
| <br>
|  
| Kulmakertoimen tarkkuusparametri
| Parameter of accuraty for slope 
|-
|-
| beta.c  
| beta.c  
| norm  
| norm  
| stokastinen
| stocastic
| (0,10^-6)  
| (0,10^-6)  
| <br>
|  
| Kulmakertoimen keskiarvo
| Mean value of slope
|-
|-
| beta[i]  
| beta[i]  
| norm  
| norm  
| stokastinen
| stocastic
| (b.c,b.t)  
| (b.c,b.t)  
| <br>
|  
| Kulmakerroin
| Slope
|-
|-
| mu[i,j]  
| mu[i,j]  
| <br>
|  
| laskettu
| calculated
| <br>
|  
| a[i]+b[i]*(x[j]-xbar)  
| a[i]+b[i]*(x[j]-xbar)  
| Pitoisuusestimaatti
| Estimation for concentration
|-
|-
| x[j]  
| x[j]  
| <br>
|  
| vakio
| constant
| <br>
|  
| annettu
| input
| Ikäluokat
| Age categories
|-
|-
| tau.c  
| tau.c  
| gamma  
| gamma  
| stokastinen
| stocastic
| 10^-3  
| 10^-3  
| <br>
|  
| Tarkkuusparametri
| Parameter of accuracy
|-
|-
| sigma  
| sigma  
| <br>
|  
| laskettu
| calculated
| <br>
|  
| 1/sqrt(tau.c)  
| 1/sqrt(tau.c)  
| Hajonta
| Dispersion term
|-
|-
| y[i,j]  
| y[i,j]  
| norm  
| norm  
| stokastinen
| stocastic
| (mu[i,j],tau.c)  
| (mu[i,j],tau.c)  
| <br>
|  
| Pitoisuusestimaatti pisteelle josta ei ole havaintoa
| Estimate for concentration without&nbsp; observation&nbsp;
|}
|}


<br>
 


Kuva 1. Picture of the model in Win BUGS<br>
Picture 1. Picture of the model in Win BUGS


<br>
[[Image:Bayes diagram.jpg|Image:Bayes_diagram.jpg]] 


http://heande.pyrkilo.fi/heande/images/f/f1/Bayes_diagram.jpg<br>
 


<br>
 
 
 


== Material  ==
== Material  ==


Datassa on silakan kokoja, ikiä sekä arvoja dioksiini-pitoisuuksille (pg/g WHOPCDD/F)
The data we used in our example consists of dioxin concentration of herring with ages, sizes and location


<br>Ajatellaan pitoisuutta kalassa, hieman samaan tyyliin, kuin rotta esimerkissä on ajateltu painoa, eli malli on hyvin samankaltainen.
 


Muuttujat:  
We create a model similar to example model in WinBUGS. Variables are analogous as demonstrated below: Variables:  


Rotan paino = Silakan pitoisuus<br>Rotan ikä = Silakan ikä<br>Monesko rotta = Mistä pyydetty
 


Tämän jälkeen WinBUGS:lla voi pyörittää rotta mallia silakka-datalla. Näin saadaan mielenkiintoiset muuttujat ja jakaumat esille.
Weight of rat = Concentration of herring Age of rat = Age of herring Number of rat (specimen #) = Location


Rotta esimerkissä siis estimoidaan jakaumat painolla jokaiselle rotalle erikseen. Meillä estimoidaan jakaumat jokaiselle pitoisuudelle alueittain. <br><br>
 


= Results  =
We ran the model with Win BUGS with herring data, and examined the distribution of variables of interest.In rat example the model estimates distributions for weight for each rat. In our model we estimate concentration distributions for each location.


== Descriptive statistics  ==
 


<br>Kuva 2. Age against concentration in herring<br><br>
== Results  ==


http://heande.pyrkilo.fi/heande/index.php/Image:Ik%C3%A4_vs_poitoisuus_silakka.jpg<br>
=== Descriptive statistics  ===


<br>
Kuva 2. Age against concentration in herring 


Kuva 3.Weight against concentration in herring<br>
[[Image:Ikä vs poitoisuus silakka.jpg|Image:Ikä_vs_poitoisuus_silakka.jpg]]     


<br>
 


http://heande.pyrkilo.fi/heande/index.php/Image:Paino_vs_pitoisuus_silakka.jpg<br>
Kuva 3.Weight against concentration in herring 


<br>
[[Image:Paino vs pitoisuus silakka.jpg|Image:Paino_vs_pitoisuus_silakka.jpg]] 


Lineaarisen regressiomallin tuloksia, joissa pitoisuutta selitetään iällä ja painolla.
Results of a linear regression model, where concentraition is explained with age and weight 


Call:<br>lm(formula = pitoisuus ~ ikä + paino, data = data)  
Call: lm(formula = pitoisuus ~ ikä + paino, data = data)  


Residuals:<br> Min 1Q Median 3Q Max <br>-7.9395 -1.3808 -0.1211 1.2573 6.4274  
Residuals: Min 1Q Median 3Q Max -7.9395 -1.3808 -0.1211 1.2573 6.4274  


Coefficients:<br> Estimate Std. Error t value Pr(&gt;|t|) <br>(Intercept) -1.68568 0.78077 -2.159 0.0363 * <br>ikä 1.66347 0.14978 11.106 2.39e-14 ***<br>paino -0.03660 0.01205 -3.038 0.0040 ** <br>---<br>Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1  
Coefficients: Estimate Std. Error t value Pr(&gt;|t|) (Intercept) -1.68568 0.78077 -2.159 0.0363 * ikä 1.66347 0.14978 11.106 2.39e-14 *** paino -0.03660 0.01205 -3.038 0.0040 ** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1  


Residual standard error: 2.575 on 44 degrees of freedom<br>Multiple R-squared: 0.7495, Adjusted R-squared: 0.7381 <br>F-statistic: 65.82 on 2 and 44 DF, p-value: 5.952e-14  
Residual standard error: 2.575 on 44 degrees of freedom Multiple R-squared: 0.7495, Adjusted R-squared: 0.7381 F-statistic: 65.82 on 2 and 44 DF, p-value: 5.952e-14  


Huomataan siis että ikä ja paino merkitseviä selittäjiä.  
Huomataan siis että ikä ja paino merkitseviä selittäjiä.  


<br>
We notice that both age and weight are significant explanators. 
 
===  Results of WinBUGS calculations  ===
 
Tables 2 and 3 represent results of our model. In Y[1,2], the first parameter denotes to concentraion of on year old herring in location one (=1), and the second parameter denotes to age category (=2). Thus, the first parameter stands for location and the second parameter for age. Some values (confidence intervall 2.5%) are less than zero (which is not plausible as concentrations can not be negative), but we can control this by not allowing values below zero in our model. In table 3 there are ages divided into four categories (including three years each) and estimates for concentrations.&nbsp; Median values do not have values below zero but 2.5% CI has negative values. Estimates in table 2 are for missing values only. Medians seem to be quite good estimates but the confidence intervalls span very wide due to lack of data.


= <br>Results of WinBUGS calculations  =
 


Seuraavissa taulukoissa 2 ja 3 esitetään hierarkinen bayes malli silakka-datalle. Huomaamisen arvoista on, että Y[1,1] on ensimmäisen alueen 1 vuotiaan silakan pitoisuus jne. Ensimmäinen indeksi on siis paikka indeksi ja toinen on ikä indeksi. Osa arvoista pienempiä kuin nolla, mutta Y:lle voi asettaa alarajaksi nolla, jolloin tästä ongelmasta päästään. Liitteet kulkukuva ja tiheydet liittyvät tähän ja aukeavat bugs:ssa. Taulukossa 3 iät on jaettu neljään ryhmään kolmen vuoden jaksoissa. 1-3, 4-6 jne. Sekä näitä vastaavat tulokset. Huomataan ettei miinuksia ole keskiarvoissa, joskin luottamusväleistä niitä löytyy. Taulukon 2 estimaatit ovat vain puuttuville arvoille. Mediaanit ovat melko hyviä arvioita, mutta luottamusvälit leviävät kohtuuttoman isoiksi, johtuen liian suuresta määrästä puuttuvia havaintoja.  
Table 2. List of results for herringdata with one year age intervals (WinBUGS).&nbsp;  [[Image:Results1.jpg]] 


<br>
 


Taulukko 2. WinBUGSin tuloslistaus silakkadatalle yhden vuoden ikä-intervallilla. <br><br>
Table 3. List of results for dioxin concentration in herring with three year age intervall (Win BUGS) 


<br>
[[Image:Results2.jpg]] 


Taulukko 3. WinBUGSin tuloslistaus silakkadatalle kolmen vuoden ikä-intervallilla. <br>
See table 4 for Original data of herring 


<br>
=== Syntaxes and iterations&nbsp;  ===


== Syntaxes and iterations&nbsp;  ==
Attached are the code of the model (model1, noodle_1 and data_1). There are age groups 1-4 and data_1 for each age categories, as well as inits (if we don't want to simulate them).


Liitteenä tulee koodit model1 , noodle_1, ja data ja data_1 joista data:ssa on 1-4 ikäluokat ja data_1 löytyy kaikille iän arvoille vastaava, sekä inits jossa alkuarvoja, mikäli niitä ei tahdota simuloida.
 


Kaikki koodit aukeavat WinBUGS ohjelmalla. Ja ovat vapaasti päivitettävissä. Ajot on tehty mallilla model1 ja datalla data ja data_1 ja todettu toimiviksi alkuarvoilla inits. Vastaavanlainen malli on esimerkeissä joka löytyy helpin alta examples vol1 Rats.  
All codes open with WinBUGS program and are updated freely. Simulations are using model1 and data_1 and they are working with inits. Similar model can be found in examples (help menu)&nbsp; vol1 Rars. Setting a limit is done by adding syntax I(0,) to the end of distribution definition.  


rajan asettaminen tapahtuu laittamalla Y:n koodin eli jakauman ja arvojen perään symbolin I(0,)
 


Ajaminen:  
Running the model:  


Mikäli ongelmia niin soita/tule kysymään. Tarvitset kolme ikkunaa auki. Modelin alta model specification, Inferencen alta samples ja modelin alta update, niin pääsee tekemään samat ajot kuin mitä on tässä tehty.  
You need to open three windows. Under model find model specification, under inference find samples, and under model find update. With these, you can run the same calculations as we just did.  


Koodit löytyy: N:\Huippuyksikko\Tutkimus\R55_PMvsDioxin\MeHgvsOmega-3\mallit\Bayes\Aiemmat ajot ja pähkäilyt
 


= <br>Conculsions  =
You can find the files from:


Bugs ohjelmistolla saatuja tuloksia voi verrata silakka-dataan, ja sen pohjalta tehdä päätelmiä.<br><br>
N:\Huippuyksikko\Tutkimus\R55_PMvsDioxin\MeHgvsOmega-3\mallit\Bayes\Aiemmat ajot ja pähkäilyt


== Example 1. ==
=== Conculsions  ===


Hierarkinen rotta esimerkki, joka löytyy winbugsin helpistä. Tähän malliin perustuu suoraan meidän käyttämät mallit.  
We can compare the results with data and draw conclusions. The median estimates seem to have pretty ok values compared the the original data.  
 
=== Rat example  ===
 
This model can be found from WinBUGS help page. Our example is based directly to this model.
 
 
 
 


Rats: a normal hierarchical model  
Rats: a normal hierarchical model  
Line 247: Line 239:
This example is taken from section 6 of Gelfand et al (1990), and concerns 30 young rats whose weights were measured weekly for five weeks. Part of the data is shown below, where Yij is the weight of the ith rat measured at age xj.  
This example is taken from section 6 of Gelfand et al (1990), and concerns 30 young rats whose weights were measured weekly for five weeks. Part of the data is shown below, where Yij is the weight of the ith rat measured at age xj.  


Weights Yij of rat i on day xj<br> xj = 8 15 22 29 36 <br> __________________________________ <br> Rat 1 151 199 246 283 320<br> Rat 2 145 199 249 293 354<br> .......<br> Rat 30 153 200 244 286 324  
 
 
Weights Y<sub>ij</sub> of rat i on day x<sub>j</sub>
 
{| cellspacing="1" cellpadding="1" border="1" style="width: 221px; height: 108px;"
|-
| x<sub>j</sub>=8  
| 15  
| 22  
| 29  
| 36  
|-
| Rat 1  
| 151  
| 199  
| 246  
| 283  
| 320  
|-
| Rat 2  
| 145  
| 199  
| 149
| 293  
| 354  
|-
| Rat 30  
| 153  
| 200  
| 244  
| 288
| 324  
|}


<br>A plot of the 30 growth curves suggests some evidence of downward curvature.  
A plot of the 30 growth curves suggests some evidence of downward curvature.  


The model is essentially a random effects linear growth curve  
The model is essentially a random effects linear growth curve  


Yij ~ Normal(i + i(xj - xbar), c)<br> <br> i ~ Normal(c, )<br> <br> i ~ Normal(c, )<br> <br>where xbar = 22, and represents the precision (1/variance) of a normal distribution. We note the absence of a parameter representing correlation between i and i unlike in Gelfand et al 1990. However, see the Birats example in Volume 2 which does explicitly model the covariance between i and i. For now, we standardise the xj's around their mean to reduce dependence between i and i in their likelihood: in fact for the full balanced data, complete independence is achieved. (Note that, in general, prior independence does not force the posterior distributions to be independent).  
Y<sub>ij</sub> ~ Normal(a<sub>i</sub> + b<sub>i</sub>(x<sub>j</sub> - x<sub>bar</sub>), T<sub>c</sub>)   a<sub>i</sub> ~ Normal(a<sub>c</sub>, T<sub>a</sub>)   b<sub>i</sub> ~ Normal(b<sub>c</sub>, T<sub>b</sub>)   where x<sub>bar</sub> = 22, and T represents the precision (1/variance) of a normal distribution. We note the absence of a parameter representing correlation between a<sub>i</sub> and b<sub>i</sub> unlike in Gelfand et al 1990. However, see the Birats example in Volume 2 which does explicitly model the covariance between a<sub>i</sub> and b<sub>i</sub>. For now, we standardise the x<sub>j</sub>'s around their mean to reduce dependence between a<sub>i</sub> and b<sub>i</sub> in their likelihood: in fact for the full balanced data, complete independence is achieved. (Note that, in general, prior independence does not force the posterior distributions to be independent).  


c , , c , , c are given independent "noninformative''priors, with two alternatives considered for  and : prior 1 is uniform on the scale of the standard deviations  = 1/sqrt() and  = 1/sqrt(), and prior 2 is a gamma(0.001, 0.001) on the precisions  and . Interest particularly focuses on the intercept at zero time (birth), denoted 0 = c - c xbar. ''  
a<sub>c</sub> , t<sub>a</sub>?, b<sub>c</sub> , T<sub>b</sub>?, T<sub>c</sub> are given independent "noninformative''priors, with two alternatives considered for T<sub>a</sub> and T<sub>b</sub>: prior 1 is uniform on the scale of the standard deviations s<sub>a</sub> = 1/sqrt(T<sub>a</sub>) and a<sub>b</sub> = 1/sqrt(T<sub>b</sub>), and prior 2 is a gamma(0.001, 0.001) on the precisions T<sub>a</sub> and T<sub>b</sub>. Interest particularly focuses on the intercept at zero time (birth), denoted a<sub>0</sub> = a<sub>c</sub> - v<sub>c </sub>x<sub>bar</sub>. ''  


Graphical model for rats example (using prior 1):  
Graphical model for rats example (using prior 1):  


<br>
 


BUGS language for rats example:  
BUGS language for rats example:  


<br>
 


<br>
 


<br>Note the use of a very flat but conjugate prior for the population effects: a locally uniform prior could also have been used.  
Note the use of a very flat but conjugate prior for the population effects: a locally uniform prior could also have been used.  


Data list(x = c(8.0, 15.0, 22.0, 29.0, 36.0), xbar = 22, N = 30, T = 5, <br> Y = structure(<br> .Data = c(151, 199, 246, 283, 320,<br> 145, 199, 249, 293, 354,<br> 147, 214, 263, 312, 328,<br> 155, 200, 237, 272, 297,<br> 135, 188, 230, 280, 323,<br> 159, 210, 252, 298, 331,<br> 141, 189, 231, 275, 305,<br> 159, 201, 248, 297, 338,<br> 177, 236, 285, 350, 376,<br> 134, 182, 220, 260, 296,<br> 160, 208, 261, 313, 352,<br> 143, 188, 220, 273, 314,<br> 154, 200, 244, 289, 325,<br> 171, 221, 270, 326, 358,<br> 163, 216, 242, 281, 312,<br> 160, 207, 248, 288, 324,<br> 142, 187, 234, 280, 316,<br> 156, 203, 243, 283, 317,<br> 157, 212, 259, 307, 336,<br> 152, 203, 246, 286, 321,<br> 154, 205, 253, 298, 334,<br> 139, 190, 225, 267, 302,<br> 146, 191, 229, 272, 302,<br> 157, 211, 250, 285, 323,<br> 132, 185, 237, 286, 331,<br> 160, 207, 257, 303, 345,<br> 169, 216, 261, 295, 333,<br> 157, 205, 248, 289, 316,<br> 137, 180, 219, 258, 291,<br> 153, 200, 244, 286, 324),<br> .Dim = c(30,5)))  
Data list(x = c(8.0, 15.0, 22.0, 29.0, 36.0), xbar = 22, N = 30, T = 5,   Y = structure( .Data = c(151, 199, 246, 283, 320, 145, 199, 249, 293, 354, 147, 214, 263, 312, 328, 155, 200, 237, 272, 297, 135, 188, 230, 280, 323, 159, 210, 252, 298, 331, 141, 189, 231, 275, 305, 159, 201, 248, 297, 338, 177, 236, 285, 350, 376, 134, 182, 220, 260, 296, 160, 208, 261, 313, 352, 143, 188, 220, 273, 314, 154, 200, 244, 289, 325, 171, 221, 270, 326, 358, 163, 216, 242, 281, 312, 160, 207, 248, 288, 324, 142, 187, 234, 280, 316, 156, 203, 243, 283, 317, 157, 212, 259, 307, 336, 152, 203, 246, 286, 321, 154, 205, 253, 298, 334, 139, 190, 225, 267, 302, 146, 191, 229, 272, 302, 157, 211, 250, 285, 323, 132, 185, 237, 286, 331, 160, 207, 257, 303, 345, 169, 216, 261, 295, 333, 157, 205, 248, 289, 316, 137, 180, 219, 258, 291, 153, 200, 244, 286, 324), .Dim = c(30,5)))  


(Note: the response data (Y) for the rats example can also be found in the file ratsy.odc in rectangular format. The covariate data (X) can be found in S-Plus format in file ratsx.odc. To load data from each of these files, focus the window containing the open data file before clicking on "Data" from the "Model" menu.)  
(Note: the response data (Y) for the rats example can also be found in the file ratsy.odc in rectangular format. The covariate data (X) can be found in S-Plus format in file ratsx.odc. To load data from each of these files, focus the window containing the open data file before clicking on "Data" from the "Model" menu.)  


Inits1 list(alpha = c(250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, <br> 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250),<br> beta = c(6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, <br> 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6),  
Inits1 list(alpha = c(250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250,   250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250), beta = c(6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,   6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6),  


<br> alpha.c = 150, beta.c = 10, <br> tau.c = 1, sigma.alpha = 1, sigma.beta = 1)  
alpha.c = 150, beta.c = 10,   tau.c = 1, sigma.alpha = 1, sigma.beta = 1)  


<br>Inits2 list(alpha = c(250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, <br> 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250),<br> beta = c(6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, <br> 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6), <br> alpha.c = 150, beta.c = 10, <br> tau.c = 1, tau.alpha = 1, tau.beta = 1)  
Inits2 list(alpha = c(250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250,   250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250), beta = c(6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6,   6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6),   alpha.c = 150, beta.c = 10,   tau.c = 1, tau.alpha = 1, tau.beta = 1)  


<br>Results  
Results  


A 1000 update burn in followed by a further 10000 updates gave the parameter estimates:  
A 1000 update burn in followed by a further 10000 updates gave the parameter estimates:  


node mean sd MC error 2.5% median 97.5% start sample<br> alpha0 106.6 3.65 0.04151 99.43 106.5 113.9 1001 10000<br> beta.c 6.185 0.1102 0.001294 5.967 6.185 6.404 1001 10000<br> sigma 6.074 0.4673 0.007724 5.247 6.044 7.068 1001 10000  
node mean sd MC error 2.5% median 97.5% start sample alpha0 106.6 3.65 0.04151 99.43 106.5 113.9 1001 10000 beta.c 6.185 0.1102 0.001294 5.967 6.185 6.404 1001 10000 sigma 6.074 0.4673 0.007724 5.247 6.044 7.068 1001 10000  


Using Prior 2 (not recommended) has negligible impact  
Using Prior 2 (not recommended) has negligible impact  


<br>
 


These results may be compared with Figure 5 of Gelfand et al 1990 --- we note that the mean gradient of independent fitted straight lines is 6.19.  
These results may be compared with Figure 5 of Gelfand et al 1990 --- we note that the mean gradient of independent fitted straight lines is 6.19.  
Line 295: Line 324:
click on one of the arrows to open the data for the missing value analysis  
click on one of the arrows to open the data for the missing value analysis  


Gelfand et al 1990 focus on the parameter estimates and the predictions for the final 4 observations on rat 26. These predictions are obtained automatically in BUGS by monitoring the relevant Y[] nodes. The following estimates were obtained:  
Gelfand et al 1990 focus on the parameter estimates and the predictions for the final 4 observations on rat 26. These predictions are obtained automatically in BUGS by monitoring the relevant Y[ ] nodes. The following estimates were obtained:  
 
We note that our estimate 6.58 of bc is substantially greater than that shown in Figure 6 of Gelfand et al 1990. However, plotting the growth curves indicates some curvature with steeper gradients at the beginning: the mean of the estimated gradients of the reduced data is 6.66, compared to 6.19 for the full data. Hence we are inclined to believe our analysis. The observed weights for rat 26 were 207, 257, 303 and 345, compared to our predictions of 204, 250, 295 and 341. 
 
Table 4. Original data for herring 
 
{| {{prettytable}}
|-
| Age
| Weight
| Concentration
|-
| 2
| 15
| 1.96
|-
| 4
| 23
| 2.49
|-
| 6
| 31
| 6.58
|-
| 7
| 43
| 13.4
|-
| 11
| 71.7
| 17.2
|-
| 2
| 17,6
| 0.876
|-
| 3
| 25.2
| 2.22
|-
| 3
| 34.1
| 1.99
|-
| 5
| 54
| 1.5
|-
| 6
| 137
| 2.18
|-
| 2
| 18
| 2
|-
| 2
| 24
| 2.52
|-
| 6
| 33
| 8.23
|-
| 9
| 52
| 15.9
|-
| 11
| 64
| 16.6
|-
| 2
| 16.3
| 0.704
|-
| 4
| 23.6
| 2.84
|-
| 7
| 34
| 6.79
|-
| 9
| 45
| 14.5
|-
| 12
| 65
| 13.6
|-
| 2
| 17
| 2.29
|-
| 3
| 24
| 3.67
|-
| 7
| 33
| 10.9
|-
| 8
| 42.9
| 11.2
|-
| 9
| 55
| 17.7
|-
| 1
| 15.3
| 0.733
|-
| 3
| 23.5
| 2.35
|-
| 5
| 30
| 4.59
|-
| 7
| 41
| 6.59
|-
| 3
| 15.8
| 2.02
|-
| 6
| 22
| 4.6
|-
| 7
| 29
| 7.4
|-
| 7
| 47
| 5.03
|-
| 6
| 144
| 7.52
|-
| 1
| 19
| 1.13
|-
| 3
| 25
| 1.186
|-
| 5
| 36
| 5.7
|-
| 2
| 13
| 1.42
|-
| 5
| 22
| 3.25
|-
| 6
| 32
| 4.44
|-
| 8
| 34
| 6.58
|-
| 10
| 106
| 3.13
|-
| 1
| 17
| 0.768
|-
| 3
| 24
| 1.79
|-
| 2
| 24
| 0.782
|-
| 3
| 48
| 1.29
|-
| 7
| 196
| 2.31
|-
|}
 
 
==== Tutkittava asia Panulle: ====
 
'''Summa PBDE pitoisuus'''
 
Luokitus ensisijaisesti rasvaprosentin mukaan (vaihtoehtoisesti ikä, paino tai pituus). Jaetaan neljään lohkoon (esim. rasvaprosentti 0-2, 2-4, 4-6, &gt;6)
 
Kokeile eri kategoriointa (rasvaprosentti, ikä, paino, pituus) mikä parhaiten selittää PBDE -pitoisuuden muuttumista.
 
Samaan tapaan kuin aiempi silakka-data. Priori-jakauma tulee kaikista PBDE-mittauksista ja rasvapidoisuuden täydentävä data tulee kustakin saman rasvapitoisuusluokan omaavista kaloista.
 
Data kirjasta Hallikainen et al. 2004. Sivut 17- 36. Kotimaisten järvi- ja merikalan dioksiinien, furaanien, dioksiinien kaltaisten PCB –yhdisteiden ja polybromattujen difenyylieettereiden pitoiuudet. EU-kalat. Elintarvikeviraston julkaisuja. Helsinki 2004.
 
==See also==
 
* [[SHELF| SHELF package for R and MATCH online tool]]
* [[:op_fi:Ven%C3%A4j%C3%A4n_vaalit_2011#Hierarkkinen_Bayes-malli_-_Hierarchial_Bayesian_model|Hierarchical Bayesian model of Russian duma election 2011]]
** [[:op_fi:Tiedosto:RussianElections2011.ods|OpenBUGS model about election]] (does not work)
* [http://www.opasnet.org/eractest/index.php/Sandbox#JAGS_test rjags test] (password required)
* [http://www.opasnet.org/eractest/index.php/Sandbox#rbugs_test rbugs test] (password required) ([http://cran.r-project.org/web/packages/rjags/index.html rjags in CRAN])
* [[:op_fi:Silakan riskiarvio|Benefit-risk assessment of Baltic herring]]
 
== References  ==
 
<references/>


We note that our estimate 6.58 of bc is substantially greater than that shown in Figure 6 of Gelfand et al 1990. However, plotting the growth curves indicates some curvature with steeper gradients at the beginning: the mean of the estimated gradients of the reduced data is 6.66, compared to 6.19 for the full data. Hence we are inclined to believe our analysis. The observed weights for rat 26 were 207, 257, 303 and 345, compared to our predictions of 204, 250, 295 and 341.<br><br>
1) Gelman Andrew, Carlin John B., Stern Hal S., and Rubin Donald B. Bayesian data analysis – second edition. Chapman &amp; Hall/CRC 2004.  
[[Category:Bayes]]

Latest revision as of 19:45, 21 July 2013


This page describes utilization of Hierarcical Bayesin method. The following page briefly introduces theory behind the method, and illustrates the use the method by examples.

The Bayesian method

The Bayes method is based on the bayesian rule. The following equations can be found from the reference 1, pages 7-8. The Bayesian rule is:

p(θ,y) = p(θ)p(y|θ)

,where p(θ) is prior distribution and on p(y|θ) is distribution acquired from the data. By contidionalizing θ with y (which is known), we get:

p(θ|y) = p(θ)p(y|θ) / p(y)

By conditionalizing this way, we obtain p(?|y), which is prior probability conditionalised by the data. This is also called posterior probability. With these equations, we can adjust presuppositions about the probability density function of the data using data itself. As an end result we obtain upgraded probability density function, which is called posterior density.

Posterior distribution ~ prior distribution x distribution estimated by the data

Hierarcical Bayesian method

We create an example of the method in this article. We study dioxin concentrations in different sea area which we assume to be share a similar form of distribution. In other words, we assume that the concentraions in different areas are samples of the total concentration distribution. Thus, we can use observations of concentrations in all areas to estimate concentrations in one particular area of interest. This means conditionalizing observated values to certain parameters. We assume probabilites to these parameters with new parameters, which are called hyperparameters.

There is an extensive explanation of the idea and the method in reference 1, pages 117-121, and you can find a picture of the method in page 119. In our example, we try to create a model which gives us a relatively accurate concentration estimate for all different areas.

Material and methods

Methods

The method used herein is called hierarcical Bayes method. Theory can be read from reference 1. We used WinBUGS and R as a modelling tool.

http://www.mrc-bsu.cam.ac.uk/bugs/

You can get the program and updates from the address above. They also include quite extensive manuals. You need to copy the text to WinBUGS to a new file, and select decode from tools menu and follow the manual. 

Table 1. Names of the variables and parameters in the model. 

Variable Distribution Nature Parameters Definition Description
alpha.tau gamma stocastic 10^-3 Parameter of accuracy
alpha.c norm stocastic (0,10^-6) Mean
alpha0 calculated (a.c)-(b.c)*xbar Starting level
alpha[i] norm stocastic (a.c,a.t) Constant term
beta.tau gamma stocastic 10^-3 Parameter of accuraty for slope
beta.c norm stocastic (0,10^-6) Mean value of slope
beta[i] norm stocastic (b.c,b.t) Slope
mu[i,j] calculated a[i]+b[i]*(x[j]-xbar) Estimation for concentration
x[j] constant input Age categories
tau.c gamma stocastic 10^-3 Parameter of accuracy
sigma calculated 1/sqrt(tau.c) Dispersion term
y[i,j] norm stocastic (mu[i,j],tau.c) Estimate for concentration without  observation 


Picture 1. Picture of the model in Win BUGS

Image:Bayes_diagram.jpg




Material

The data we used in our example consists of dioxin concentration of herring with ages, sizes and location


We create a model similar to example model in WinBUGS. Variables are analogous as demonstrated below: Variables:


Weight of rat = Concentration of herring Age of rat = Age of herring Number of rat (specimen #) = Location


We ran the model with Win BUGS with herring data, and examined the distribution of variables of interest.In rat example the model estimates distributions for weight for each rat. In our model we estimate concentration distributions for each location.


Results

Descriptive statistics

Kuva 2. Age against concentration in herring  

Image:Ikä_vs_poitoisuus_silakka.jpg


Kuva 3.Weight against concentration in herring

Image:Paino_vs_pitoisuus_silakka.jpg

Results of a linear regression model, where concentraition is explained with age and weight

Call: lm(formula = pitoisuus ~ ikä + paino, data = data)

Residuals: Min 1Q Median 3Q Max -7.9395 -1.3808 -0.1211 1.2573 6.4274

Coefficients: Estimate Std. Error t value Pr(>|t|) (Intercept) -1.68568 0.78077 -2.159 0.0363 * ikä 1.66347 0.14978 11.106 2.39e-14 *** paino -0.03660 0.01205 -3.038 0.0040 ** --- Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 2.575 on 44 degrees of freedom Multiple R-squared: 0.7495, Adjusted R-squared: 0.7381 F-statistic: 65.82 on 2 and 44 DF, p-value: 5.952e-14

Huomataan siis että ikä ja paino merkitseviä selittäjiä.

We notice that both age and weight are significant explanators.

Results of WinBUGS calculations

Tables 2 and 3 represent results of our model. In Y[1,2], the first parameter denotes to concentraion of on year old herring in location one (=1), and the second parameter denotes to age category (=2). Thus, the first parameter stands for location and the second parameter for age. Some values (confidence intervall 2.5%) are less than zero (which is not plausible as concentrations can not be negative), but we can control this by not allowing values below zero in our model. In table 3 there are ages divided into four categories (including three years each) and estimates for concentrations.  Median values do not have values below zero but 2.5% CI has negative values. Estimates in table 2 are for missing values only. Medians seem to be quite good estimates but the confidence intervalls span very wide due to lack of data.


Table 2. List of results for herringdata with one year age intervals (WinBUGS). 


Table 3. List of results for dioxin concentration in herring with three year age intervall (Win BUGS)

See table 4 for Original data of herring

Syntaxes and iterations 

Attached are the code of the model (model1, noodle_1 and data_1). There are age groups 1-4 and data_1 for each age categories, as well as inits (if we don't want to simulate them).


All codes open with WinBUGS program and are updated freely. Simulations are using model1 and data_1 and they are working with inits. Similar model can be found in examples (help menu)  vol1 Rars. Setting a limit is done by adding syntax I(0,) to the end of distribution definition.


Running the model:

You need to open three windows. Under model find model specification, under inference find samples, and under model find update. With these, you can run the same calculations as we just did.


You can find the files from:

N:\Huippuyksikko\Tutkimus\R55_PMvsDioxin\MeHgvsOmega-3\mallit\Bayes\Aiemmat ajot ja pähkäilyt

Conculsions

We can compare the results with data and draw conclusions. The median estimates seem to have pretty ok values compared the the original data.

Rat example

This model can be found from WinBUGS help page. Our example is based directly to this model.



Rats: a normal hierarchical model

This example is taken from section 6 of Gelfand et al (1990), and concerns 30 young rats whose weights were measured weekly for five weeks. Part of the data is shown below, where Yij is the weight of the ith rat measured at age xj.


Weights Yij of rat i on day xj

xj=8 15 22 29 36
Rat 1 151 199 246 283 320
Rat 2 145 199 149 293 354
Rat 30 153 200 244 288 324
A plot of the 30 growth curves suggests some evidence of downward curvature. 

The model is essentially a random effects linear growth curve

Yij ~ Normal(ai + bi(xj - xbar), Tc) ai ~ Normal(ac, Ta) bi ~ Normal(bc, Tb) where xbar = 22, and T represents the precision (1/variance) of a normal distribution. We note the absence of a parameter representing correlation between ai and bi unlike in Gelfand et al 1990. However, see the Birats example in Volume 2 which does explicitly model the covariance between ai and bi. For now, we standardise the xj's around their mean to reduce dependence between ai and bi in their likelihood: in fact for the full balanced data, complete independence is achieved. (Note that, in general, prior independence does not force the posterior distributions to be independent).

ac , ta?, bc , Tb?, Tc are given independent "noninformativepriors, with two alternatives considered for Ta and Tb: prior 1 is uniform on the scale of the standard deviations sa = 1/sqrt(Ta) and ab = 1/sqrt(Tb), and prior 2 is a gamma(0.001, 0.001) on the precisions Ta and Tb. Interest particularly focuses on the intercept at zero time (birth), denoted a0 = ac - vc xbar.

Graphical model for rats example (using prior 1):


BUGS language for rats example:



Note the use of a very flat but conjugate prior for the population effects: a locally uniform prior could also have been used.

Data list(x = c(8.0, 15.0, 22.0, 29.0, 36.0), xbar = 22, N = 30, T = 5, Y = structure( .Data = c(151, 199, 246, 283, 320, 145, 199, 249, 293, 354, 147, 214, 263, 312, 328, 155, 200, 237, 272, 297, 135, 188, 230, 280, 323, 159, 210, 252, 298, 331, 141, 189, 231, 275, 305, 159, 201, 248, 297, 338, 177, 236, 285, 350, 376, 134, 182, 220, 260, 296, 160, 208, 261, 313, 352, 143, 188, 220, 273, 314, 154, 200, 244, 289, 325, 171, 221, 270, 326, 358, 163, 216, 242, 281, 312, 160, 207, 248, 288, 324, 142, 187, 234, 280, 316, 156, 203, 243, 283, 317, 157, 212, 259, 307, 336, 152, 203, 246, 286, 321, 154, 205, 253, 298, 334, 139, 190, 225, 267, 302, 146, 191, 229, 272, 302, 157, 211, 250, 285, 323, 132, 185, 237, 286, 331, 160, 207, 257, 303, 345, 169, 216, 261, 295, 333, 157, 205, 248, 289, 316, 137, 180, 219, 258, 291, 153, 200, 244, 286, 324), .Dim = c(30,5)))

(Note: the response data (Y) for the rats example can also be found in the file ratsy.odc in rectangular format. The covariate data (X) can be found in S-Plus format in file ratsx.odc. To load data from each of these files, focus the window containing the open data file before clicking on "Data" from the "Model" menu.)

Inits1 list(alpha = c(250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250), beta = c(6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6),

alpha.c = 150, beta.c = 10, tau.c = 1, sigma.alpha = 1, sigma.beta = 1)

Inits2 list(alpha = c(250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250, 250), beta = c(6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6, 6), alpha.c = 150, beta.c = 10, tau.c = 1, tau.alpha = 1, tau.beta = 1)

Results

A 1000 update burn in followed by a further 10000 updates gave the parameter estimates:

node mean sd MC error 2.5% median 97.5% start sample alpha0 106.6 3.65 0.04151 99.43 106.5 113.9 1001 10000 beta.c 6.185 0.1102 0.001294 5.967 6.185 6.404 1001 10000 sigma 6.074 0.4673 0.007724 5.247 6.044 7.068 1001 10000

Using Prior 2 (not recommended) has negligible impact


These results may be compared with Figure 5 of Gelfand et al 1990 --- we note that the mean gradient of independent fitted straight lines is 6.19.

Gelfand et al 1990 also consider the problem of missing data, and delete the last observation of cases 6-10, the last two from 11-20, the last 3 from 21-25 and the last 4 from 26-30. The appropriate data file is obtained by simply replacing data values by NA (see below). The model specification is unchanged, since the distinction between observed and unobserved quantities is made in the data file and not the model specification.

click on one of the arrows to open the data for the missing value analysis

Gelfand et al 1990 focus on the parameter estimates and the predictions for the final 4 observations on rat 26. These predictions are obtained automatically in BUGS by monitoring the relevant Y[ ] nodes. The following estimates were obtained:

We note that our estimate 6.58 of bc is substantially greater than that shown in Figure 6 of Gelfand et al 1990. However, plotting the growth curves indicates some curvature with steeper gradients at the beginning: the mean of the estimated gradients of the reduced data is 6.66, compared to 6.19 for the full data. Hence we are inclined to believe our analysis. The observed weights for rat 26 were 207, 257, 303 and 345, compared to our predictions of 204, 250, 295 and 341.

Table 4. Original data for herring

Age Weight Concentration
2 15 1.96
4 23 2.49
6 31 6.58
7 43 13.4
11 71.7 17.2
2 17,6 0.876
3 25.2 2.22
3 34.1 1.99
5 54 1.5
6 137 2.18
2 18 2
2 24 2.52
6 33 8.23
9 52 15.9
11 64 16.6
2 16.3 0.704
4 23.6 2.84
7 34 6.79
9 45 14.5
12 65 13.6
2 17 2.29
3 24 3.67
7 33 10.9
8 42.9 11.2
9 55 17.7
1 15.3 0.733
3 23.5 2.35
5 30 4.59
7 41 6.59
3 15.8 2.02
6 22 4.6
7 29 7.4
7 47 5.03
6 144 7.52
1 19 1.13
3 25 1.186
5 36 5.7
2 13 1.42
5 22 3.25
6 32 4.44
8 34 6.58
10 106 3.13
1 17 0.768
3 24 1.79
2 24 0.782
3 48 1.29
7 196 2.31


Tutkittava asia Panulle:

Summa PBDE pitoisuus

Luokitus ensisijaisesti rasvaprosentin mukaan (vaihtoehtoisesti ikä, paino tai pituus). Jaetaan neljään lohkoon (esim. rasvaprosentti 0-2, 2-4, 4-6, >6)

Kokeile eri kategoriointa (rasvaprosentti, ikä, paino, pituus) mikä parhaiten selittää PBDE -pitoisuuden muuttumista.

Samaan tapaan kuin aiempi silakka-data. Priori-jakauma tulee kaikista PBDE-mittauksista ja rasvapidoisuuden täydentävä data tulee kustakin saman rasvapitoisuusluokan omaavista kaloista.

Data kirjasta Hallikainen et al. 2004. Sivut 17- 36. Kotimaisten järvi- ja merikalan dioksiinien, furaanien, dioksiinien kaltaisten PCB –yhdisteiden ja polybromattujen difenyylieettereiden pitoiuudet. EU-kalat. Elintarvikeviraston julkaisuja. Helsinki 2004.

See also

References


1) Gelman Andrew, Carlin John B., Stern Hal S., and Rubin Donald B. Bayesian data analysis – second edition. Chapman & Hall/CRC 2004.