Subsetting data in plot

[kevindavenport]

In R I can plot additional points based on some other criteria using the R subset command as follows:

%%R -i DF_diff_xy # list object to be transferred to python here
install.packages("ggplot2") # Had to add this for some reason, shouldn't be necessary
library(ggplot2)
df = data.frame(DF_diff_xy)
plot = ggplot(df, aes(x = X, y = Y)) + 
geom_point(alpha = .8, color = 'dodgerblue',size = 5) +
geom_point(data=subset(df, Y >= 6.7 | X >= 4), color = 'red',size = 6) +
theme(axis.text.x = element_text(size= rel(1.5),angle=90, hjust=1)) +
ggtitle('Distance Pairs with outliers highlighted in red')
print(plot)

In Python my thinking was I could specify a row slice of a dataframe for additional highlight as so:

from ggplot import *

ggplot(DF_diff_xy, aes(x = 'X', y ='Y')) + \
    geom_point(alpha=1, size=100, color='dodgerblue') + \
    geom_point(data = DF_diff_xy[:1], alpha=1, color='black')

This didn't work however, any ideas?

Thanks, Kevin Davenport http://kldavenport.com

[jankatins]

Currently it is not possible to specify data per geom.

@glamp Currently the ggplot._get_layers() method is not really the equivalent to to ggplots2 "layer": the real layer information is inggplot.geoms. But as the data is set inggplot._get_layers(...)` and the geoms are iterated afterwards, the geom (real "Layer") can't set it's own dataset. I would suggest change the iteration to

for geom in geoms:
    _data = geom.data or self.data
    for sub_layer in self._get_layers(_data):
         [...]

geom._init__() would then pop data from the args and save it to geom.data.

[jankatins]

Ok, it's (of course... :-/ ) not as easy: when you do that (and transform the data from the geom with the aes like ggplot.__init__() does... geom specific aes mapping was also not implemented yet), then there is an error because the plotting code assumes that there are some "assigned colors", but as this is a new dataset, they aren't... So actually this also needs to look into how to assign colors and so on...

One way would be to refactor the assign_*(gg) functions to build_*_mapping(data, aes, legend, gg), which would set the needed columns in data based on the passed in aes and gg (only manual color mapping and so on...). But that's for tomorrow...

[jankatins]

This can be closed

[kevindavenport]

Awesome Jan, thank you for your contribution. I think I can update http://kldavenport.com/mahalanobis-distance-and-outliers/ now :)

[jankatins]

Let's see if it works for you :-)

[jankatins]

downloaded you ipynb and run it here: it works :-)

[jankatins]

Just for the reference, here are the changes I had to do to the :

# needed because in latest pandas, the series are not anymore numpy arrays...
# see https://github.com/pydata/pandas/issues/5698
xydata = DF_diff_xy.values
xycols = DF_diff_xy.columns
--
%%R -i xydata,xycols # list object to be transferred to python here
install.packages("ggplot2") # Had to add this for some reason, shouldn't be necessary
library(ggplot2)
df = data.frame(xydata)
names(df) <- c(xycols)
plot = ggplot(df, aes(x = X, y = Y)) + 
geom_point(alpha = .8, color = 'dodgerblue',size = 5) +
geom_point(data=subset(df, Y >= 6.7 | X >= 4), color = 'red',size = 6) +
theme(axis.text.x = element_text(size= rel(1.5),angle=90, hjust=1)) +
ggtitle('Distance Pairs with outliers highlighted in red')
print(plot)
--
from ggplot import *

ggplot(DF_diff_xy, aes(x = 'X', y ='Y')) + \
    geom_point(alpha=1, size=100, color='dodgerblue') + \
    geom_point(data = DF_diff_xy[(DF_diff_xy.Y >= 6.7) | (DF_diff_xy.X >= 4)],alpha=1, size = 100, color='red')  

[kevindavenport]

Just tried it, works perfectly! Will start updating my blog post now :)