{smcl}
{* *! opl_plot_best, v5, GCerulli, 11aug2025}
{title:Title}
 
{pstd}
{cmd:opl_plot_best} {hline 2} Plot observed versus the
maximal expected reward, observed versus optimal treatment.

{marker syntax}{...}{title:Syntax}

{p 8 17 2}
{cmd:opl_plot_best} {it:varlist} [{cmd:,}
{opt gr_reward_train(name)}
{opt gr_action_train(name)}]

were:

{pstd}
{it:varlist} must be an ordered list of {bf:four} variables:

{p 8 12 2}
(1) the expected outcome (based on the training policy);{p_end}
{p 8 12 2}
(2) the observed treatment (i.e., the training policy);{p_end}
{p 8 12 2}
(3) the maximal expected outcome (based on the best policy);{p_end}
{p 8 12 2}
(4) the best treatment (i.e., the optimal policy).{p_end}

{marker description}{title:Description}

{pstd}
After running {cmd:opl_best_treat}, the command {cmd:opl_plot_best} plots observed versus the maximal expected reward, and observed versus optimal treatment, using the variables generated by {cmd:opl_best_treat}.
{p_end} 

{marker options}{title:Options}

{phang}
{opt gr_reward_train(name)} saves the reward plot (observed outcome vs. maximal expected outcome)
to a graph named {it:name} in the current working directory.

{phang}
{opt gr_action_train(name)} saves the action plot (observed treatment vs. optimal treatment)
to a graph named {it:name} in the current working directory.

{marker remarks}{title:Remarks}

{pstd}
The command assumes that the four variables in {it:varlist} appear in the exact order listed above.
Results are typically generated by {cmd:opl_best_treat} and then passed to {cmd:opl_plot_best}.
{p_end} 

{marker example}{title:Examples}

{phang}* After {cmd:opl_best_treat} has produced the needed variables:{p_end}
{phang}{cmd:.opl_plot_best} y obs_treat y_max t_best{p_end}

{phang}* Save both graphs:{p_end}
{phang}{cmd:.opl_plot_best} y obs_treat y_max t_best, gr_reward_train(reward_train) gr_action_train(action_train){p_end}

{dlgtab:Author}

{phang}Giovanni Cerulli{p_end}
{phang}IRCrES-CNR{p_end}
{phang}Research Institute for Sustainable Economic Growth, National Research Council of Italy{p_end}
{phang}E-mail: {browse "mailto:giovanni.cerulli@cnr.it":giovanni.cerulli@cnr.it}{p_end}

{dlgtab:Also see:}

{psee}
{help opl_ma_fb}, {help opl_ma_vf}, {help opl_best_treat}{p_end}