The best way to visualize complex data

Question

I use the data below to create a graph in R using ggplot2.

Hour.of.day Model N Distance.travelled sd se ci 1 0100 h300_fv30 60 3.6264709 5.078277 0.6556027 1.3118579 2 0100 h300_fv35 60 2.9746019 5.313252 0.6859379 1.3725586 3 0100 h300_fv40 60 3.0422525 3.950650 0.5100267 1.0205610 4 0200 h300_fv30 60 4.3323896 6.866003 0.8863972 1.7736767 5 0200 h300_fv35 60 3.5567420 6.259378 0.8080823 1.6169689 6 0200 h300_fv40 60 2.5232512 4.533234 0.5852380 1.1710585 7 0300 h300_fv30 60 3.1800537 5.303506 0.6846797 1.3700409 8 0300 h300_fv35 60 2.9281442 4.445953 0.5739700 1.1485113 9 0300 h300_fv40 60 2.5078045 4.058295 0.5239236 1.0483687 10 0400 h300_fv30 60 3.3408231 4.567161 0.5896180 1.1798229 11 0400 h300_fv35 60 2.8679676 5.396700 0.6967110 1.3941155 12 0400 h300_fv40 60 3.1615813 4.244155 0.5479180 1.0963815 13 0500 h300_fv30 60 3.8117851 6.970900 0.8999394 1.8007745 14 0500 h300_fv35 60 2.1130581 3.925906 0.5068323 1.0141691 15 0500 h300_fv40 60 3.6430531 4.905484 0.6332953 1.2672209 16 0600 h300_fv30 60 3.5234762 5.150027 0.6648657 1.3303931 17 0600 h300_fv35 60 2.0341804 3.192176 0.4121082 0.8246266 18 0600 h300_fv40 60 3.2838958 3.770624 0.4867855 0.9740555 19 0700 h300_fv30 60 3.8327926 6.521022 0.8418603 1.6845587 20 0700 h300_fv35 60 1.6933289 2.607322 0.3366039 0.6735428 21 0700 h300_fv40 60 2.3896956 3.435656 0.4435413 0.8875241 22 0800 h300_fv30 60 3.3077466 6.504371 0.8397107 1.6802573 23 0800 h300_fv35 60 1.4823307 3.556884 0.4591917 0.9188405 24 0800 h300_fv40 60 2.4161741 3.571444 0.4610715 0.9226019 25 0900 h300_fv30 60 2.1506438 2.893029 0.3734885 0.7473487 26 0900 h300_fv35 60 1.8821961 3.457929 0.4464167 0.8932778 27 0900 h300_fv40 60 1.7896335 2.714514 0.3504423 0.7012334 28 1000 h300_fv30 60 2.5107475 5.491835 0.7089929 1.4186914 29 1000 h300_fv35 60 0.9491365 2.061712 0.2661658 0.5325966 30 1000 h300_fv40 60 1.6678013 3.234033 0.4175119 0.8354393 31 1100 h300_fv30 60 1.8602186 3.365695 0.4345093 0.8694511 32 1100 h300_fv35 60 1.4385708 2.869765 0.3704851 0.7413389 33 1100 h300_fv40 60 1.1273899 2.010280 0.2595261 0.5193105 34 1200 h300_fv30 60 1.4870763 2.112841 0.2727667 0.5458048 35 1200 h300_fv35 60 2.5295481 4.740384 0.6119810 1.2245711 36 1200 h300_fv40 60 1.6551202 3.051420 0.3939366 0.7882653 37 1300 h300_fv30 60 2.8791490 4.925870 0.6359271 1.2724872 38 1300 h300_fv35 60 2.4731563 5.266690 0.6799268 1.3605303 39 1300 h300_fv40 60 4.5989133 8.394460 1.0837201 2.1685189 40 1400 h300_fv30 60 1.5050205 3.188480 0.4116310 0.8236717 41 1400 h300_fv35 60 1.7615688 3.064842 0.3956693 0.7917325 42 1400 h300_fv40 60 2.2766514 5.215937 0.6733746 1.3474194 43 1500 h300_fv30 60 1.9097882 2.770040 0.3576106 0.7155772 44 1500 h300_fv35 60 2.0109347 4.070014 0.5254365 1.0513961 45 1500 h300_fv40 60 1.6316881 4.119681 0.5318485 1.0642264 46 1600 h300_fv30 60 3.3246263 5.352698 0.6910304 1.3827486 47 1600 h300_fv35 60 2.0389703 3.781869 0.4882372 0.9769604 48 1600 h300_fv40 60 1.0204568 2.205685 0.2847527 0.5697888 49 1700 h300_fv30 60 3.6132519 5.467875 0.7058996 1.4125019 50 1700 h300_fv35 60 2.1139255 4.178283 0.5394140 1.0793648 51 1700 h300_fv40 60 1.5547818 3.411135 0.4403756 0.8811895 52 1800 h300_fv30 60 5.0552532 7.344069 0.9481152 1.8971742 53 1800 h300_fv35 60 2.1832792 3.824244 0.4937078 0.9879070 54 1800 h300_fv40 60 1.6532516 3.273697 0.4226325 0.8456856 55 1900 h300_fv30 60 5.6107731 6.891023 0.8896272 1.7801399 56 1900 h300_fv35 60 2.9822004 5.958244 0.7692060 1.5391777 57 1900 h300_fv40 60 2.7111394 3.798765 0.4904184 0.9813250 58 2000 h300_fv30 60 6.0438385 7.126952 0.9200855 1.8410868 59 2000 h300_fv35 60 3.9517888 6.462761 0.8343388 1.6695081 60 2000 h300_fv40 60 3.9508503 5.374253 0.6938130 1.3883167 61 2100 h300_fv30 60 4.2144712 5.648673 0.7292406 1.4592070 62 2100 h300_fv35 60 2.2205186 3.397391 0.4386013 0.8776392 63 2100 h300_fv40 60 3.9000010 5.881409 0.7592866 1.5193290 64 2200 h300_fv30 60 3.9478958 5.584154 0.7209112 1.4425401 65 2200 h300_fv35 60 3.1612149 4.788883 0.6182421 1.2370996 66 2200 h300_fv40 60 3.7812992 6.424478 0.8293965 1.6596186 67 2300 h300_fv30 61 3.3860628 5.176299 0.6627571 1.3257117 68 2300 h300_fv35 61 3.7427743 6.257596 0.8012031 1.6026448 69 2300 h300_fv40 61 3.6674335 4.945831 0.6332487 1.2666861 70 2400 h300_fv30 59 3.8745470 5.763821 0.7503856 1.5020600 71 2400 h300_fv35 59 3.1284346 5.016476 0.6530895 1.3073007 72 2400 h300_fv40 59 3.7563017 4.819053 0.6273872 1.2558520 

Graph function

 ggplot(my_data, aes(x=Hour.of.day, y=Distance.travelled, colour=Model)) + geom_errorbar(aes(ymin = Distance.travelled - ci, ymax = Distance.travelled + ci), width=.1, position=position_dodge(2)) + geom_line(position=position_dodge(2)) + geom_point(position=position_dodge(2)) + scale_x_discrete(breaks=c("0600", "1200", "1800", "2400")) + theme(axis.ticks = element_blank()) 

Differentiation of three separate patterns is difficult to do in the resulting graph.

Does anyone have any suggestions on ways to improve visualization so that the three separate templates can be better differentiated? For example, is there some way to emphasize midpoints and set confidence intervals in the background?

Answer 1

Use lines and ribbons:

 library(ggplot2) ggplot(my_data, aes(x=Hour.of.day, y=Distance.travelled, fill=Model)) + theme_bw()+ geom_line(aes(colour=Model))+ geom_ribbon(aes(ymin = Distance.travelled - ci, ymax = Distance.travelled + ci),alpha=0.4)+ scale_x_discrete(breaks=c("0600", "1200", "1800", "2400")) + theme(axis.ticks = element_blank()) ggsave("ribbonplot.png",width=7,height=4) 

You can make the lines wider ( lwd ) or the ribbons weaker ( alpha ) if you want to emphasize the average value pattern more.

Answer 2

Here's another way using faces:

 ggplot(gg,aes(x=Hour.of.day, y=Distance.travelled)) + geom_pointrange(aes(ymin=Distance.travelled-ci,ymax=Distance.travelled+ci,color=Model))+ facet_grid(Model~.) + stat_smooth(formula=y~1, method="lm",linetype=2,se=F)+ geom_abline(aes(slope=0,intercept=mean(Distance.travelled)),linetype=3) 

The main idea here is that the data should have a reference system (here the average value of the distance traveled for this model). This tells you at a glance when the distance traveled is significantly different from the average. The gray dashed line is the average value for all models, which indicates that this model has a tendency to travel more or less across all models.

If you set se = T in the stat_smooth(...) call, you will also get variability in the means, but anything that obscures, I think, distracts from the main point.