Boxplot problem of flat line

[Samantha Will]

Hello,

* Example generated by -dataex-. For more info, type help dataex
clear
input float(fundnr year mf_m_cf)
 444 2009     .9111743
1606 2003    1.8259066
3484 2010     9.627364
 743 2016     .4432602
3691 2011     4.885561
1935 2015   -1.8404514
 444 1993    .01043051
 984 1993 -.0044249827
2524 1996     .9388815
 610 1992            .
1417 1995    10.157868
 444 2015     .9147166
 527 1990            .
 767 2007      31.3497
 444 1991   -.10612977
 444 1999     .6728972
 402 2014     -.705271
1219 1999   -4.4375644
 444 2005   -2.2521157
 444 1992      4.22313
 560 2004    -.3882031
3580 2009   -.20625106
 527 1991     6.328609
 979 1996    -5.362143
1060 2001    -3.597908
 417 2002   -1.1781628
 811 2008    -1.620225
 444 1991     .1711222
 767 2009   -1.3755553
1669 2001   -.14327554
2708 2003     .9402666
3347 2008    .13227251
2708 2004            .
 444 2004   -2.2446256
1146 2004    .08893927
3248 2006     9.885372
 444 1999     .3944024
3396 2007     3.144457
3705 2017    12.843843
3460 1994            .
 335 2008     -.675634
 579 2003   -2.0698159
 940 2010     1.912574
2708 2003     17.15932
2708 2003   -.05014902
3079 2012    1.4607174
2708 2004            .
 453 2008     .4533027
 862 2014   -1.2433937
 444 2002    -2.389034
 650 2013    -.2421079
1117 2017    .12475158
1417 1993            .
2793 2013    -.9765097
1023 2017   -1.9454733
3487 2013    -2.230105
 799 1994   .002836363
 993 2015    -1.622088
1320 2015     .3555873
 439 1995     2.656444
2994 2010    .04025678
3553 2014     .1185691
1064 1995    -.6680014
3296 2003     .9756665
1925 2012    .25422385
 444 2001     4.793575
 778 2015     .9230105
 995 1993     4.540443
 444 2006   -1.2819324
 612 1994     .9295382
3250 2011    -3.901114
3487 2013    1.1448343
2413 2005     5.719396
 918 1996    1.2646472
 444 2003   -1.8748924
1246 2018   -.25754893
 865 1994    -.3397882
3386 2018     -.932922
3487 2008     3.253638
 869 2009    -.3591767
 995 2003    -.2239342
1608 2000            .
1104 2007   -1.0762625
 630 2005   -1.1473552
2735 2007     3.981054
 353 2000     1.657574
3487 2009     9.839337
1671 1996     9.620604
 650 2016    -1.475633
 444 1994     2.024926
 444 1994    -.3641732
 444 1995   -1.7175412
1868 2002     -17.2737
2809 2010   -2.0511134
1082 2004     1.545442
3386 2014    1.5415833
 428 2012    -.4344161
3306 2015   -4.4798098
 829 2007    1.4708675
 350 2013   -.21686088
end

I am trying to find the outliers through boxplot, but there is no box when I run 'graph boxplot mf_m_cf', the outliers have a prominent effect on my average flows, which leads to a much higher result than intended. So I can see the outliers but there is a flat line of a box.

[Nick Cox]

quantile mf_m_cf, ms(oh)

will give you a much richer appreciation of the distribution.

As a step further in the same direction I used qplot (Stata Journal) and transplot (SSC) on your data example with the original data and the so-called neglog transformation sign(y) * ln(1 + |y|)) which behaves like y near 0, like ln y for y >> 0 and like --ln(-y)) for y << 0.

transplot qplot mf_m_cf, trans(@   sign(@)*log1p(abs(@)))

The "outliers" don't look that extraordinary -- in this data example, viewed on a reasonable scale.

[Samantha Will]

quantile mf_m_cf, ms(oh)

will give you a much richer appreciation of the distribution.

As a step further in the same direction I used qplot (Stata Journal) and transplot (SSC) on your data example with the original data and the so-called neglog transformation sign(y) * ln(1 + |y|)) which behaves like y near 0, like ln y for y >> 0 and like --ln(-y)) for y << 0.

transplot qplot mf_m_cf, trans(@ sign(@)*log1p(abs(@)))

[ATTACH=CONFIG]n1657199[/ATTACH]

The "outliers" don't look that extraordinary -- in this data example, viewed on a reasonable scale.

Thank you, the cashflows go up to 2.50e+07 but this occurs only once which is not shown in my example dataset