AttributeError: 'str' object has no attribute 'view' in Seaborn , Scatterplot

user8270077 picture user8270077 · Jan 3, 2019 · Viewed 8.6k times · Source

I am getting a bizarre exception in Seaborn.

For a reproducible example:

toy_data.to_json()
'{"X":{"0":0.12765045,"1":0.0244816152,"2":0.1263715245,"3":0.0246376768,"4":0.1108581319,"5":0.1406719382,"6":0.1358105564,"7":0.1245863432,"8":0.1175445352,"9":0.1188479018,"10":0.1113148159,"11":0.117455495,"12":0.110555662,"13":0.1328567106,"14":0.103064284,"15":0.1119474442,"16":0.119390455,"17":0.1246727756,"18":0.1117827155,"19":0.1169972547},"Y":{"0":0.1241083714,"1":0.1394242378,"2":0.1225010796,"3":0.0077080173,"4":0.1198371354,"5":0.0029026989,"6":0.1259473297,"7":0.0,"8":0.0,"9":0.1214620231,"10":0.1204110739,"11":0.0,"12":0.1194570059,"13":0.0014971676,"14":0.1184584731,"15":0.1212061305,"16":0.1221438778,"17":0.0,"18":0.1209991075,"19":0.0},"Label":{"0":"17","1":"3","2":"17","3":"0","4":"14","5":"21","6":"16","7":"23","8":"20","9":"15","10":"14","11":"20","12":"14","13":"22","14":"13","15":"14","16":"15","17":"23","18":"14","19":"20"},"Probability":{"0":1.0,"1":1.0,"2":1.0,"3":1.0,"4":1.0,"5":1.0,"6":1.0,"7":1.0,"8":1.0,"9":1.0,"10":1.0,"11":1.0,"12":1.0,"13":1.0,"14":1.0,"15":1.0,"16":1.0,"17":1.0,"18":0.9101796407,"19":1.0}}'

toy_data.head()
       X           Y    Label   Probability
0   0.127650    0.124108    17  1.0
1   0.024482    0.139424    3   1.0
2   0.126372    0.122501    17  1.0
3   0.024638    0.007708    0   1.0
4   0.110858    0.119837    14  1.0

sns.scatterplot(x = toy_data.X, y = toy_data.Y, hue = toy_data.Label.values, alpha = 0.5)

AttributeError: 'str' object has no attribute 'view'

Similar exception with this syntax:

sns.scatterplot(x = 'X', y = 'Y', data = toy_data, hue = 'Label', alpha = 0.5)

Answer

Prageeth Jayathissa picture Prageeth Jayathissa · Nov 2, 2019

This is a peculiar issue with seaborn where the color palette needs to equal the number of labels

sns.scatterplot(x = toy_data.X, y = toy_data.Y, hue = toy_data.Label, alpha = 0.5,
                palette=sns.color_palette("Set1", toy_data.Label.nunique()) )

If not, seaborn will apply numerical functions to map the number of hue categories to the default palette of four colours. This is why you get an error when the values are a string