diff options
| author | Johannes Ranke <jranke@uni-bremen.de> | 2019-11-01 15:34:28 +0100 |
|---|---|---|
| committer | Johannes Ranke <jranke@uni-bremen.de> | 2019-11-01 15:34:28 +0100 |
| commit | ce73c044b949154e3bc3e715b9b79e1360b3f794 (patch) | |
| tree | 28f477a3142f1efa463a8d8569f924b9064e8637 /docs/reference/confint.mkinfit.html | |
| parent | e7c65ee913d4a84da0957d2ebb89abfbc444de56 (diff) | |
Make the 'quadratic' the default for 'confint'
Also the documentation was improved here and there
Diffstat (limited to 'docs/reference/confint.mkinfit.html')
| -rw-r--r-- | docs/reference/confint.mkinfit.html | 109 |
1 files changed, 64 insertions, 45 deletions
diff --git a/docs/reference/confint.mkinfit.html b/docs/reference/confint.mkinfit.html index 54696ff5..6015fbf6 100644 --- a/docs/reference/confint.mkinfit.html +++ b/docs/reference/confint.mkinfit.html @@ -36,9 +36,14 @@ <meta property="og:title" content="Confidence intervals for parameters of mkinfit objects — confint.mkinfit" /> -<meta property="og:description" content="The default method 'profile' is based on the profile likelihood for each -parameter. The method uses two nested optimisations. The speed of the method -could likely be improved by using the method of Venzon and Moolgavkar (1988)." /> +<meta property="og:description" content="The default method 'quadratic' is based on the quadratic approximation of +the curvature of the likelihood function at the maximum likelihood parameter +estimates. +The alternative method 'profile' is based on the profile likelihood for each +parameter. The method uses two nested optimisations and can take a very long +time, even if parallelized by specifying 'cores' on unixoid platforms. The +speed of the method could likely be improved by using the method of Venzon +and Moolgavkar (1988)." /> <meta name="twitter:card" content="summary" /> @@ -71,7 +76,7 @@ could likely be improved by using the method of Venzon and Moolgavkar (1988)." / </button> <span class="navbar-brand"> <a class="navbar-link" href="../index.html">mkin</a> - <span class="version label label-default" data-toggle="tooltip" data-placement="bottom" title="Released version">0.9.49.6</span> + <span class="version label label-default" data-toggle="tooltip" data-placement="bottom" title="Released version">0.9.49.8</span> </span> </div> @@ -135,14 +140,19 @@ could likely be improved by using the method of Venzon and Moolgavkar (1988)." / </div> <div class="ref-description"> - <p>The default method 'profile' is based on the profile likelihood for each -parameter. The method uses two nested optimisations. The speed of the method -could likely be improved by using the method of Venzon and Moolgavkar (1988).</p> + <p>The default method 'quadratic' is based on the quadratic approximation of +the curvature of the likelihood function at the maximum likelihood parameter +estimates. +The alternative method 'profile' is based on the profile likelihood for each +parameter. The method uses two nested optimisations and can take a very long +time, even if parallelized by specifying 'cores' on unixoid platforms. The +speed of the method could likely be improved by using the method of Venzon +and Moolgavkar (1988).</p> </div> <pre class="usage"><span class='co'># S3 method for mkinfit</span> <span class='fu'><a href='https://rdrr.io/r/stats/confint.html'>confint</a></span>(<span class='no'>object</span>, <span class='no'>parm</span>, <span class='kw'>level</span> <span class='kw'>=</span> <span class='fl'>0.95</span>, <span class='kw'>alpha</span> <span class='kw'>=</span> <span class='fl'>1</span> - - <span class='no'>level</span>, <span class='no'>cutoff</span>, <span class='kw'>method</span> <span class='kw'>=</span> <span class='fu'><a href='https://rdrr.io/r/base/c.html'>c</a></span>(<span class='st'>"profile"</span>, <span class='st'>"quadratic"</span>), + <span class='no'>level</span>, <span class='no'>cutoff</span>, <span class='kw'>method</span> <span class='kw'>=</span> <span class='fu'><a href='https://rdrr.io/r/base/c.html'>c</a></span>(<span class='st'>"quadratic"</span>, <span class='st'>"profile"</span>), <span class='kw'>transformed</span> <span class='kw'>=</span> <span class='fl'>TRUE</span>, <span class='kw'>backtransform</span> <span class='kw'>=</span> <span class='fl'>TRUE</span>, <span class='kw'>cores</span> <span class='kw'>=</span> <span class='fu'><a href='https://rdrr.io/r/base/Round.html'>round</a></span>(<span class='fu'>detectCores</span>()/<span class='fl'>2</span>), <span class='kw'>quiet</span> <span class='kw'>=</span> <span class='fl'>FALSE</span>, <span class='no'>...</span>)</pre> @@ -174,11 +184,11 @@ cutoff value overrides arguments 'level' and 'alpha'</p></td> </tr> <tr> <th>method</th> - <td><p>The 'profile' method searches the parameter space for the -cutoff of the confidence intervals by means of a likelihood ratio test. -The 'quadratic' method approximates the likelihood function at the -optimised parameters using the second term of the Taylor expansion, using -a second derivative (hessian) contained in the object.</p></td> + <td><p>The 'quadratic' method approximates the likelihood function at +the optimised parameters using the second term of the Taylor expansion, +using a second derivative (hessian) contained in the object. +The 'profile' method searches the parameter space for the +cutoff of the confidence intervals by means of a likelihood ratio test.</p></td> </tr> <tr> <th>transformed</th> @@ -231,20 +241,27 @@ machines, cores > 1 is not supported.</p></td> #> parent_0 73.0641834 92.1392181 #> k_parent_sink 0.2170293 0.4235348 #> sigma 3.1307772 8.0628314</div><div class='input'> +<span class='co'># Set the number of cores for the profiling method for further examples</span> +<span class='kw'>if</span> (<span class='fu'><a href='https://rdrr.io/r/base/identical.html'>identical</a></span>(<span class='fu'><a href='https://rdrr.io/r/base/Sys.getenv.html'>Sys.getenv</a></span>(<span class='st'>"NOT_CRAN"</span>), <span class='st'>"true"</span>)) { + <span class='no'>n_cores</span> <span class='kw'><-</span> <span class='kw pkg'>parallel</span><span class='kw ns'>::</span><span class='fu'><a href='https://rdrr.io/r/parallel/detectCores.html'>detectCores</a></span>() - <span class='fl'>1</span> +} <span class='kw'>else</span> { + <span class='no'>n_cores</span> <span class='kw'><-</span> <span class='fl'>1</span> +} +<span class='kw'>if</span> (<span class='fu'><a href='https://rdrr.io/r/base/Sys.getenv.html'>Sys.getenv</a></span>(<span class='st'>"TRAVIS"</span>) <span class='kw'>!=</span> <span class='st'>""</span>) <span class='no'>n_cores</span> <span class='kw'>=</span> <span class='fl'>1</span> +<span class='kw'>if</span> (<span class='fu'><a href='https://rdrr.io/r/base/Sys.info.html'>Sys.info</a></span>()[<span class='st'>"sysname"</span>] <span class='kw'>==</span> <span class='st'>"Windows"</span>) <span class='no'>n_cores</span> <span class='kw'>=</span> <span class='fl'>1</span> + <span class='no'>SFO_SFO</span> <span class='kw'><-</span> <span class='fu'><a href='mkinmod.html'>mkinmod</a></span>(<span class='kw'>parent</span> <span class='kw'>=</span> <span class='fu'><a href='mkinsub.html'>mkinsub</a></span>(<span class='st'>"SFO"</span>, <span class='st'>"m1"</span>), <span class='kw'>m1</span> <span class='kw'>=</span> <span class='fu'><a href='mkinsub.html'>mkinsub</a></span>(<span class='st'>"SFO"</span>), <span class='kw'>quiet</span> <span class='kw'>=</span> <span class='fl'>TRUE</span>) <span class='no'>SFO_SFO.ff</span> <span class='kw'><-</span> <span class='fu'><a href='mkinmod.html'>mkinmod</a></span>(<span class='kw'>parent</span> <span class='kw'>=</span> <span class='fu'><a href='mkinsub.html'>mkinsub</a></span>(<span class='st'>"SFO"</span>, <span class='st'>"m1"</span>), <span class='kw'>m1</span> <span class='kw'>=</span> <span class='fu'><a href='mkinsub.html'>mkinsub</a></span>(<span class='st'>"SFO"</span>), <span class='kw'>use_of_ff</span> <span class='kw'>=</span> <span class='st'>"max"</span>, <span class='kw'>quiet</span> <span class='kw'>=</span> <span class='fl'>TRUE</span>) <span class='no'>f_d_1</span> <span class='kw'><-</span> <span class='fu'><a href='mkinfit.html'>mkinfit</a></span>(<span class='no'>SFO_SFO</span>, <span class='fu'><a href='https://rdrr.io/r/base/subset.html'>subset</a></span>(<span class='no'>FOCUS_2006_D</span>, <span class='no'>value</span> <span class='kw'>!=</span> <span class='fl'>0</span>), <span class='kw'>quiet</span> <span class='kw'>=</span> <span class='fl'>TRUE</span>) -<span class='fu'><a href='https://rdrr.io/r/base/system.time.html'>system.time</a></span>(<span class='no'>ci_profile</span> <span class='kw'><-</span> <span class='fu'><a href='https://rdrr.io/r/stats/confint.html'>confint</a></span>(<span class='no'>f_d_1</span>, <span class='kw'>cores</span> <span class='kw'>=</span> <span class='fl'>1</span>, <span class='kw'>quiet</span> <span class='kw'>=</span> <span class='fl'>TRUE</span>))</div><div class='output co'>#> User System verstrichen -#> 51.646 0.000 51.673 </div><div class='input'><span class='co'># The following does not save much time, as parent_0 takes up most of the time</span> -<span class='co'># system.time(ci_profile <- confint(f_d_1, cores = 5))</span> -<span class='co'># system.time(ci_profile <- confint(f_d_1,</span> -<span class='co'># c("k_parent_sink", "k_parent_m1", "k_m1_sink", "sigma"), cores = 1))</span> -<span class='co'># If we exclude parent_0 (the confidence of which is often of minor interest), we get a nice</span> -<span class='co'># performance improvement from about 30 seconds to about 12 seconds</span> -<span class='co'># system.time(ci_profile_no_parent_0 <- confint(f_d_1, </span> -<span class='co'># c("k_parent_sink", "k_parent_m1", "k_m1_sink", "sigma"), cores = 4))</span> -<span class='no'>ci_profile</span></div><div class='output co'>#> 2.5% 97.5% +<span class='fu'><a href='https://rdrr.io/r/base/system.time.html'>system.time</a></span>(<span class='no'>ci_profile</span> <span class='kw'><-</span> <span class='fu'><a href='https://rdrr.io/r/stats/confint.html'>confint</a></span>(<span class='no'>f_d_1</span>, <span class='kw'>method</span> <span class='kw'>=</span> <span class='st'>"profile"</span>, <span class='kw'>cores</span> <span class='kw'>=</span> <span class='fl'>1</span>, <span class='kw'>quiet</span> <span class='kw'>=</span> <span class='fl'>TRUE</span>))</div><div class='output co'>#> User System verstrichen +#> 51.297 0.000 51.328 </div><div class='input'><span class='co'># Using more cores does not save much time here, as parent_0 takes up most of the time</span> +<span class='co'># If we additionally exclude parent_0 (the confidence of which is often of</span> +<span class='co'># minor interest), we get a nice performance improvement from about 50</span> +<span class='co'># seconds to about 12 seconds if we use at least four cores</span> +<span class='fu'><a href='https://rdrr.io/r/base/system.time.html'>system.time</a></span>(<span class='no'>ci_profile_no_parent_0</span> <span class='kw'><-</span> <span class='fu'><a href='https://rdrr.io/r/stats/confint.html'>confint</a></span>(<span class='no'>f_d_1</span>, <span class='kw'>method</span> <span class='kw'>=</span> <span class='st'>"profile"</span>, + <span class='fu'><a href='https://rdrr.io/r/base/c.html'>c</a></span>(<span class='st'>"k_parent_sink"</span>, <span class='st'>"k_parent_m1"</span>, <span class='st'>"k_m1_sink"</span>, <span class='st'>"sigma"</span>), <span class='kw'>cores</span> <span class='kw'>=</span> <span class='no'>n_cores</span>))</div><div class='output co'>#> <span class='message'>Profiling the likelihood</span></div><div class='output co'>#> User System verstrichen +#> 0.006 0.003 11.435 </div><div class='input'><span class='no'>ci_profile</span></div><div class='output co'>#> 2.5% 97.5% #> parent_0 96.456003650 1.027703e+02 #> k_parent_sink 0.040762501 5.549764e-02 #> k_parent_m1 0.046786482 5.500879e-02 @@ -267,29 +284,26 @@ machines, cores > 1 is not supported.</p></td> <span class='co'># interval based on the untransformed fit for k_m1_sink</span> <span class='no'>rel_diffs_transformed</span> <span class='kw'><-</span> <span class='fu'><a href='https://rdrr.io/r/base/MathFun.html'>abs</a></span>((<span class='no'>ci_quadratic_transformed</span> - <span class='no'>ci_profile</span>)/<span class='no'>ci_profile</span>) <span class='no'>rel_diffs_untransformed</span> <span class='kw'><-</span> <span class='fu'><a href='https://rdrr.io/r/base/MathFun.html'>abs</a></span>((<span class='no'>ci_quadratic_untransformed</span> - <span class='no'>ci_profile</span>)/<span class='no'>ci_profile</span>) -<span class='no'>rel_diffs_transformed</span></div><div class='output co'>#> 2.5% 97.5% -#> parent_0 0.0005407854 0.0002218012 -#> k_parent_sink 0.0066452394 0.0083795930 -#> k_parent_m1 0.0001833903 0.0020092090 -#> k_m1_sink 0.0307278240 0.0290580487 -#> sigma 0.0550252516 0.0327066836</div><div class='input'><span class='no'>rel_diffs_untransformed</span></div><div class='output co'>#> 2.5% 97.5% -#> parent_0 0.0005407854 0.0002218011 -#> k_parent_sink 0.0067996407 0.0025717594 -#> k_parent_m1 0.0037382781 0.0011843074 -#> k_m1_sink 0.0146745610 0.0025299672 -#> sigma 0.0550252516 0.0327066836</div><div class='input'> -<span class='co'># Set the number of cores for further examples</span> -<span class='kw'>if</span> (<span class='fu'><a href='https://rdrr.io/r/base/identical.html'>identical</a></span>(<span class='fu'><a href='https://rdrr.io/r/base/Sys.getenv.html'>Sys.getenv</a></span>(<span class='st'>"NOT_CRAN"</span>), <span class='st'>"true"</span>)) { - <span class='no'>n_cores</span> <span class='kw'><-</span> <span class='kw pkg'>parallel</span><span class='kw ns'>::</span><span class='fu'><a href='https://rdrr.io/r/parallel/detectCores.html'>detectCores</a></span>() - <span class='fl'>1</span> -} <span class='kw'>else</span> { - <span class='no'>n_cores</span> <span class='kw'><-</span> <span class='fl'>1</span> -} -<span class='kw'>if</span> (<span class='fu'><a href='https://rdrr.io/r/base/Sys.getenv.html'>Sys.getenv</a></span>(<span class='st'>"TRAVIS"</span>) <span class='kw'>!=</span> <span class='st'>""</span>) <span class='no'>n_cores</span> <span class='kw'>=</span> <span class='fl'>1</span> -<span class='kw'>if</span> (<span class='fu'><a href='https://rdrr.io/r/base/Sys.info.html'>Sys.info</a></span>()[<span class='st'>"sysname"</span>] <span class='kw'>==</span> <span class='st'>"Windows"</span>) <span class='no'>n_cores</span> <span class='kw'>=</span> <span class='fl'>1</span> +<span class='no'>rel_diffs_transformed</span> <span class='kw'><</span> <span class='no'>rel_diffs_untransformed</span></div><div class='output co'>#> 2.5% 97.5% +#> parent_0 FALSE FALSE +#> k_parent_sink TRUE FALSE +#> k_parent_m1 TRUE FALSE +#> k_m1_sink FALSE FALSE +#> sigma FALSE FALSE</div><div class='input'><span class='fu'><a href='https://rdrr.io/r/base/Round.html'>signif</a></span>(<span class='no'>rel_diffs_transformed</span>, <span class='fl'>3</span>)</div><div class='output co'>#> 2.5% 97.5% +#> parent_0 0.000541 0.000222 +#> k_parent_sink 0.006650 0.008380 +#> k_parent_m1 0.000183 0.002010 +#> k_m1_sink 0.030700 0.029100 +#> sigma 0.055000 0.032700</div><div class='input'><span class='fu'><a href='https://rdrr.io/r/base/Round.html'>signif</a></span>(<span class='no'>rel_diffs_untransformed</span>, <span class='fl'>3</span>)</div><div class='output co'>#> 2.5% 97.5% +#> parent_0 0.000541 0.000222 +#> k_parent_sink 0.006800 0.002570 +#> k_parent_m1 0.003740 0.001180 +#> k_m1_sink 0.014700 0.002530 +#> sigma 0.055000 0.032700</div><div class='input'> <span class='co'># Investigate a case with formation fractions</span> <span class='no'>f_d_2</span> <span class='kw'><-</span> <span class='fu'><a href='mkinfit.html'>mkinfit</a></span>(<span class='no'>SFO_SFO.ff</span>, <span class='fu'><a href='https://rdrr.io/r/base/subset.html'>subset</a></span>(<span class='no'>FOCUS_2006_D</span>, <span class='no'>value</span> <span class='kw'>!=</span> <span class='fl'>0</span>), <span class='kw'>quiet</span> <span class='kw'>=</span> <span class='fl'>TRUE</span>) -<span class='no'>ci_profile_ff</span> <span class='kw'><-</span> <span class='fu'><a href='https://rdrr.io/r/stats/confint.html'>confint</a></span>(<span class='no'>f_d_2</span>, <span class='kw'>cores</span> <span class='kw'>=</span> <span class='no'>n_cores</span>)</div><div class='output co'>#> <span class='message'>Profiling the likelihood</span></div><div class='input'><span class='no'>ci_profile_ff</span></div><div class='output co'>#> 2.5% 97.5% +<span class='no'>ci_profile_ff</span> <span class='kw'><-</span> <span class='fu'><a href='https://rdrr.io/r/stats/confint.html'>confint</a></span>(<span class='no'>f_d_2</span>, <span class='kw'>method</span> <span class='kw'>=</span> <span class='st'>"profile"</span>, <span class='kw'>cores</span> <span class='kw'>=</span> <span class='no'>n_cores</span>)</div><div class='output co'>#> <span class='message'>Profiling the likelihood</span></div><div class='input'><span class='no'>ci_profile_ff</span></div><div class='output co'>#> 2.5% 97.5% #> parent_0 96.456003650 1.027703e+02 #> k_parent 0.090911032 1.071578e-01 #> k_m1 0.003892605 6.702778e-03 @@ -310,9 +324,14 @@ machines, cores > 1 is not supported.</p></td> <span class='no'>rel_diffs_untransformed_ff</span> <span class='kw'><-</span> <span class='fu'><a href='https://rdrr.io/r/base/MathFun.html'>abs</a></span>((<span class='no'>ci_quadratic_untransformed_ff</span> - <span class='no'>ci_profile_ff</span>)/<span class='no'>ci_profile_ff</span>) <span class='co'># While the confidence interval for the parent rate constant is closer to</span> <span class='co'># the profile based interval when using the internal parameter</span> -<span class='co'># transformation, the intervals for the other parameters are 'better</span> +<span class='co'># transformation, the interval for the metabolite rate constant is 'better</span> <span class='co'># without internal parameter transformation.</span> -<span class='no'>rel_diffs_transformed_ff</span></div><div class='output co'>#> 2.5% 97.5% +<span class='no'>rel_diffs_transformed_ff</span> <span class='kw'><</span> <span class='no'>rel_diffs_untransformed_ff</span></div><div class='output co'>#> 2.5% 97.5% +#> parent_0 TRUE TRUE +#> k_parent TRUE TRUE +#> k_m1 FALSE FALSE +#> f_parent_to_m1 TRUE FALSE +#> sigma FALSE TRUE</div><div class='input'><span class='no'>rel_diffs_transformed_ff</span></div><div class='output co'>#> 2.5% 97.5% #> parent_0 0.0005408012 0.0002217857 #> k_parent 0.0009596303 0.0009003981 #> k_m1 0.0307277425 0.0290579163 |