ValueError: s must be a scalar, or the same size as x and y

Question

xs=np.array([2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25])
ys=np.array([10,12,20,22,21,25,30,21,32,34,35,30,50,45,55,60,66,64,67,72,74,80,79,84])

plt.figure(figsize=(8,6))
sns.scatterplot(x=xs,y=ys,marker='o',s=ys*25,color='g',alpha=0.5)
plt.title('scatter plot')
plt.xlabel('xs value')
plt.ylabel('ys value')
plt.show()

I wanted to draw a bubble plot. I created a bubble plot before using this same code. But I recently reinstalled all anaconda and when I use this code I keep getting the error ValueError: s must be a scalar, or the same size as x and y

Answer 1

Code works fine for me.

• python 3.7.6
• pandas 1.0.3

In short, using s in a plot requires that either (a) you use a scalar (single number), or (b) the length of it match the length of x and y, so each plotted point can be assigned it's own size.

The text "must be a scalar, or the same size" is intended to suggest that, but may seem cryptic.

Since others may easily encounter this error, I wanted to document the common cause and its fix here.

I encountered this error doing something like this:

df[some_filter].plot.scatter(x='x', y='y', s=df['s'])

The reason is that the x and y fields are using a subset of df (per filter on the left) but the scale parameter s does not use that filtered DataFrame. Therefore the length of s does not match the length of x & y, and scatter does not know how to size all the points.

If you encounter this situation, it may be easiest to assign the filtered DataFrame to a new variable, and then use that variable for both the scatter() call and the s parameter.

Example:

temp = df[some_filter]
temp.plot.scatter(x='x', y='y', s=temp['s'])

Answer 2

I had this issue when I was trying to use a filtered column of a dataframe. So, I solved it with something like this:

import numpy as np
test_b=SOLEUR.loc[SOLEUR['type']=='buy']['cost']
y_b = test_b.astype(np.float)
plt.scatter(....s=y_b)