Recursion capacity

M3. Hyperparameter tuning M3. Analysis of hyper-parameter search results
#1

Hello,

I was wondering what recursion capacity is needed to run the parallel coordinates plot?

#2

What do you define as recursion capacity?

#3

When I tried to replicate the code to produce the px.parallel_coordinates graph in my own Notebook the following error occurred: “maximum recursion depth exceeded while calling a Python object”. To find out what my recursion limit was I wrote:
import sys
print(sys.getrecursionlimit())
And I found that my limit was 1000. The Python interpreter limits the depths of recursion to help avoiding infinite recursions, resulting in stack overflows. But when I tried to extend it for calculation using:
sys.setrecursionlimit(10000), my computer crashed.

#4

Recursion errors are usually linked to a bug rather than an intended recursive call :slight_smile:
Could you provide the snippet of code that trigger this error such that we can investigate?

#5

Sure, the snippet of the code is the following:

import numpy as np
import plotly.express as px

fig = px.parallel_coordinates(
    cv_results.rename(shorten_param, axis=1).apply(
        {
            "learning_rate": np.log10,
            "max_leaf_nodes": np.log2,
            "max_bins": np.log2,
            "min_samples_leaf": np.log10,
            "l2_regularization": np.log10,
            "mean_test_score": lambda x: x,
        }
    ),
    color="mean_test_score",
    color_continuous_scale=px.colors.sequential.Viridis,
)
fig.show()
#6

This is unexpected. I cannot reproduce this problem.

On the MOOC platform, we use plotly 5.6.0 and pandas 1.3.3.

Maybe this is a bug with a specific plotly or pandas version?

import plotly, pandas

plotly.__version__, pandas.__version__

Could you please include a link to a copy of the full traceback you get posted on https://gists.github.com for instance?