sheet-cell-classifier

RandomForest cell role classifier for Excel spreadsheets, used by sheet-call-tree.

Model

Predicts whether a spreadsheet cell is a header (0) or data (1). Trained on CTC (CIUS + SAUS) and ENTRANT datasets.

Features (23): gap-proximity (dist_above, dist_left), row/col numeric fractions, format fields (bold, italic, colors, borders, alignment, data type), value type flags.

Usage

from huggingface_hub import hf_hub_download
import joblib

model_path = hf_hub_download(repo_id="roksechs/sheet-cell-classifier", filename="cell_classifier.joblib")
clf = joblib.load(model_path)

sheet-call-tree >= 0.1.2 downloads and uses this model automatically.