Hello,
I just tried the matplotlib-0.83.2 and found a problem in
hist(normed=1). Sometimes the *normalized* y values can be greater
than 1.0. I looked into the source code and find the code does not
perform the supposed *normalization*.
Here's the code that shows the problem on a very unbalanced data:
data = [3]*1000
data.append(4)
n, bins, patches = hist(data, 30, normed=1)
print n
I got something like:
[ 29.97002997, 0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0. , 0. ,
0. , 0. , 0. , 0.02997003,]
the first column has a value much higher than 1.0, it's not a
probability distribution I believe.
Zhang Le