将Pandas DataFrame转换为h2o.H2OFrame时出现AttributeError：'DataFrame'对象没有'as_matrix'属性

Ah, I’ve dealt with this exact headache before—let’s get this sorted out for you.

What’s causing the error? #

Pandas deprecated the as_matrix() method way back in version 0.23, and completely removed it starting with Pandas 1.0.0. If you’re seeing this error, it’s either:

• You’re explicitly using as_matrix() in your own code, or
• Your installed version of H2O is outdated and still relies on this removed Pandas method.

Solutions to fix it #

1. If you’re calling as_matrix() directly in your code

Replace as_matrix() with Pandas’ official replacement: to_numpy(). Here’s how to adjust your conversion code:

# Old code that triggers the error
h2o_frame = h2o.H2OFrame(df.as_matrix())

# Fixed code using to_numpy()
h2o_frame = h2o.H2OFrame(df.to_numpy())

Even better—you don’t even need to convert to a numpy array manually! H2O can handle Pandas DataFrames directly:

# Even simpler: pass the DataFrame directly
h2o_frame = h2o.H2OFrame(df)

2. If the error comes from H2O’s internal code

If you didn’t write as_matrix() yourself, your H2O version is too old to work with modern Pandas. Here’s how to fix this:

• Upgrade H2O to the latest stable version (this is the recommended long-term fix). Use pip or conda to update:

  # Using pip
  pip install --upgrade h2o
  
  # Using conda
  conda update h2o
  

  After upgrading, restart your Python session to ensure the new version loads properly.
• Temporary fix: Downgrade Pandas (only if upgrading H2O isn’t an option). Install a Pandas version that still supports as_matrix() (e.g., 0.25.x):

  pip install pandas==0.25.3
  

Quick check to verify #

After making changes, run a quick test to confirm the conversion works:

import pandas as pd
import h2o

# Initialize H2O
h2o.init()

# Create a sample DataFrame
df = pd.DataFrame({"col1": [1,2,3], "col2": [4,5,6]})

# Convert to H2OFrame
h2o_df = h2o.H2OFrame(df)
print(h2o_df)

内容的提问来源于stack exchange，提问作者Divyansh Sharma