Unable to make axis logarithmic in 3D plot

Question

I am trying to make a 3D plot using matplotlib inside jupyter-notebook. I am using a dataset from kaggle.

The schema is the following

LotArea	SalePrice	YrSold	PoolArea
8450	208500	2008	0
9600	181500	2007	0
...	...	...	...

When I plot with linear axes, everything is OK:

import matplotlib.pyplot as plt
fig = plt.figure(figsize=(10, 15))
ax = plt.axes(projection='3d')

area_data = dataset_chosen["LotArea"]
price_data = dataset_chosen["SalePrice"]
year_data = dataset_chosen["YrSold"]

cmhot = plt.get_cmap("hot")

ax.scatter3D(xs=area_data, ys=price_data, zs=year_data, c=dataset_chosen["PoolArea"])

#ax.set_xscale("log")

ax.set_xlabel("Area")
ax.set_ylabel("Price")
ax.set_zlabel("Year")

plt.show()

And when I try to make x scale logarithmic (uncomment #ax.set_xscale("log")), the plot does not look like a plot.

How to make X scale logarithmic?

Answer 1

If you check here, there is a discussion on the same. This is a limitation/bug within 3d plots. As mentioned there, there is a workaround... basically, you need to manually do that scaling. Below is the updated code to do that. Hope this is what you are looking for... Note I used log10 as the numbers align up nicely.

dataset_chosen=pd.read_csv('train.csv')
fig = plt.figure(figsize=(10, 15))
ax = plt.axes(projection='3d')

area_data = np.log10(dataset_chosen["LotArea"])  ## Changed to LOG-10
price_data = dataset_chosen["SalePrice"]
year_data = dataset_chosen["YrSold"]

cmhot = plt.get_cmap("hot")

ax.scatter3D(xs=area_data, ys=price_data, zs=year_data, c=dataset_chosen["PoolArea"])

## Set the xticks and xticklables to what you want it to be...
xticks=[100,1000,10000,100000]
ax.set_xticks(np.log10(xticks))
ax.set_xticklabels(xticks)

#ax.set_xscale('log')
ax.set_xlabel("Area")
ax.set_ylabel("Price")
ax.set_zlabel("Year")

plt.show()