Fit graph to window on y-axis

[Paul Shlapa]

Dear Stata Experts,

I have a following code, which produces graphs that are being saved in .tif format. However on y-axis the graphs sometimes are too wide.
Does anyone have an idea, why?

capture noisily twoway ///
                (scatter MM_TP TP [aw = total_SalesRevenue_EUR_per_Year] if groups_TP == `i' & PricingCategory_short == "`v'" & TP <= `m_max', title("`Identifier_graph' `v'", color(maroon)) ytitle(Material Margin (%)) mcolor(navy) msymbol(circle_hollow) leg(off)) ///
                (scatter MM_TP TP if groups_TP_TP == `i' & total_SalesRevenue_EUR_per_Year == 0 & PricingCategory_short == "`v'" & TP <= `m_max', msize(tiny) mcolor(navy) msymbol(circle)) ///
                (${function_`v'}, range (`m_1_`v'' `m_2_`v'') lpattern(dash) lcolor(white)) || (${function_`v'_1}, range (`m_min' `m_1_`v'') lpattern(dash) lcolor(white)) || (${function_`v'_2}, range (`m_2_`v'' `m_max') lpattern(dash) lcolor(white)) ///
                (${function_`a'}, range (`m_1_`a'' `m_2_`a'') lcolor(maroon)) || (${function_`a'_1}, range (`m_min' `m_1_`a'') lcolor(maroon)) || (${function_`a'_2}, range (`m_2_`a'' `m_max') lcolor(maroon)) ///
                (${function_`b'}, range (`m_1_`b'' `m_2_`b'') lcolor(maroon)) || (${function_`b'_1}, range (`m_min' `m_1_`b'') lcolor(maroon)) || (${function_`b'_2}, range (`m_2_`b'' `m_max') lcolor(maroon))
                
        
                graph export "${directory}/Output Data/Main regression graphs/MM/`g'/`v'/g_`Identifier_graph'.tif", replace

The lines are some global macros (${function_`...'} in form y = ....

Some output is more convenient, but still a lot of useless space being shown.

I understand that one can define y-axis range, but still why are the graphs by default so wide (starting from 40 to 100, instead of around 70 to 100)?

Thank you for any support in this topic.

Sincerely,

Pavlo

[Nick Cox]

Sometimes a graph is drawn to respect the ranges of variables even if the extreme observations concerned are not shown on the graph. Your code is not fully explicit about all your choices, but I can't see that you are explicitly setting the y axis range or labels, which is why you should be puzzled. A way to experiment is temporarily to drop all observations not needed for the graph.

[Paul Shlapa]

Well i have also tried to manually define y-axis range an fit the graph into it (from 0 to 100), but that did not work either.

capture noisily twoway ///
        (scatter MM_$main_regressor TP [aw = total_SalesRevenue_EUR_per_Year] if groups_${main_regressor}_${main_regressor} == `i', title("`Identifier_graph'", color(maroon)) ytitle(Material Margin (%)) mcolor(navy) msymbol(circle_hollow)) ///
        (scatter MM_$main_regressor TP if groups_${main_regressor}_${main_regressor} == `i' & total_SalesRevenue_EUR_per_Year == 0, msize(tiny) mcolor(navy) msymbol(circle)) ///
        (scatter MM_TP TP [aw = total_SalesRevenue_EUR_per_Year] if groups_${main_regressor}_${main_regressor} == `i', mcolor(red) msymbol(X)) ///
        (scatter MM_TP TP if groups_${main_regressor}_${main_regressor} == `i', msize(tiny) mcolor(red) msymbol(circle)) ///
        (${function_`a'}, range (`m_1_`a'' `m_2_`a'') lcolor(maroon)) || (${function_`a'_1}, range (`m_min' `m_1_`a'') lcolor(maroon)) || (${function_`a'_2}, range (`m_2_`a'' `m_max') lcolor(maroon)) ///
        (${function_`b'}, range (`m_1_`b'' `m_2_`b'') lcolor(maroon)) || (${function_`b'_1}, range (`m_min' `m_1_`b'') lcolor(maroon)) || (${function_`b'_2}, range (`m_2_`b'' `m_max') lcolor(maroon)) ///
        (${function_`v'}, range (`m_1_`v'' `m_2_`v'') lpattern(dash) lcolor(white)) || (${function_`v'_1}, range (`m_min' `m_1_`v'') lpattern(dash) lcolor(white)) || (${function_`v'_2}, range (`m_2_`v'' `m_max') lpattern(dash) lcolor(white)), ///
        yscale(range(0 100)) legend(off)

yrange also does not work.



Without the Lines everything works.

capture noisily twoway ///
        (scatter MM_$main_regressor TP [aw = total_SalesRevenue_EUR_per_Year] if groups_${main_regressor}_${main_regressor} == `i', title("`Identifier_graph'", color(maroon)) ytitle(Material Margin (%)) mcolor(navy) msymbol(circle_hollow)) ///
        (scatter MM_$main_regressor TP if groups_${main_regressor}_${main_regressor} == `i' & total_SalesRevenue_EUR_per_Year == 0, msize(tiny) mcolor(navy) msymbol(circle)) ///
        (scatter MM_TP TP [aw = total_SalesRevenue_EUR_per_Year] if groups_${main_regressor}_${main_regressor} == `i', mcolor(red) msymbol(X)) ///
        (scatter MM_TP TP if groups_${main_regressor}_${main_regressor} == `i', msize(tiny) mcolor(red) msymbol(circle))

Any ideas, what else to try?

[Paul Shlapa]

For some reason, when i exclude lines, that go from minimum observation on x-axis to the same value (basically range(min min)), the graph is being fully shown.

But i still do not understand why any line, irrespective of its range (here 350 350) should affect the graph output in the way as shown at the beginning of this post?

capture noisily twoway ///
        (scatter MM_$main_regressor TP [aw = total_SalesRevenue_EUR_per_Year] if groups_${main_regressor}_${main_regressor} == `i', title("`Identifier_graph'", color(maroon)) ytitle(Material Margin (%)) xtitle(TP (€)) mcolor(navy) msymbol(circle_hollow)) ///
        (scatter MM_$main_regressor TP if groups_${main_regressor}_${main_regressor} == `i' & total_SalesRevenue_EUR_per_Year == 0, msize(tiny) mcolor(navy) msymbol(circle)) ///
        (scatter MM_TP TP [aw = total_SalesRevenue_EUR_per_Year] if groups_${main_regressor}_${main_regressor} == `i', mcolor(red) msymbol(X)) ///
        (scatter MM_TP TP if groups_${main_regressor}_${main_regressor} == `i', msize(tiny) mcolor(red) msymbol(circle)) ///
        (${function_`a'}, range (`m_1_`a'' `m_2_`a'') lcolor(maroon)) || (${function_`a'_2}, range (`m_2_`a'' `m_max') lcolor(maroon)) ///
        (${function_`b'}, range (`m_1_`b'' `m_2_`b'') lcolor(maroon)) || (${function_`b'_2}, range (`m_2_`b'' `m_max') lcolor(maroon)) ///
        (${function_`v'}, range (`m_1_`v'' `m_2_`v'') lpattern(dash) lcolor(white)) || (${function_`v'_2}, range (`m_2_`v'' `m_max') lpattern(dash) lcolor(white)), ///
        legend(off)

[Nick Cox]

It is hard to help further as

we do not have access to your data

we have no way of knowing earlier definitions of local and global macros

(in my case) I can't read through code like that and easily filter out what is secondary and what is problematic without a lot of experiment

In addition, you appear to be using some non-default scheme, which may be relevant.

If you want to pursue this further we need please a self-contained reproducible example so that we can copy code from here and follow on our copy of Stata