Series.iteritems() removed in pandas 2.x causing crash in split_tt

[bleedblack1]

The split_tt function in utils.py uses Series.iteritems(), which was deprecated in pandas 1.5.0 and removed in pandas 2.0.0.

Since pyproject.toml does not define an upper bound for the pandas dependency, fresh installations pull pandas 2.x, resulting in the following error:

AttributeError: 'Series' object has no attribute 'iteritems'

This causes a complete crash when using dataset splitting or model evaluation utilities.

---

Steps to Reproduce

pip install trapiche  # installs pandas>=2.0

import pandas as pd
from trapiche.utils import split_tt

df = pd.DataFrame({
    'SAMPLE_ID': ['s1','s2'],
    'project': ['p1','p1'],
    'lineage': ['root:A:B','root:A:C'],
    'max_depth': [3,3]
})

split_tt(df, 0.2, 42, 'root:A')

---

Expected Behavior

The function executes successfully and returns a DataFrame with an IS_TEST column.

---

Actual Behavior

AttributeError: 'Series' object has no attribute 'iteritems'

---

Impact

• Dataset splitting fails completely
• Model retraining and evaluation workflows are blocked
• Fresh installs are unusable with pandas 2.x

---

Proposed Fix

Replace:

.iteritems()

with:

.items()

.items() is supported in all pandas versions ≥1.0 and works as a direct replacement.

---

I’d be happy to submit a PR with this fix if it aligns with your expectations.

[bleedblack1]

@SantiagoSanchezF Would it be okay if I try to fix this bug, since it’s currently assigned to you?

[SantiagoSanchezF]

Hi Nishant @bleedblack1
Many thanks for your proposed changes. Yes, a PR would be great 🙏