modules/utils.py

import numpy as np
import scipy.io as sio

def load_mat_as_numeric(path, x_key="X", y_key="Y"):
    data = sio.loadmat(path)
    X_raw = data[x_key]
    y_raw = data[y_key]

    # Step 1: flatten MATLAB cell array elements
    def clean_cell_array(arr):
        cleaned = []
        for row in arr:
            new_row = []
            for elem in row:
                # elem is usually array(['46.0'])
                if isinstance(elem, np.ndarray):
                    elem = elem[0]  # '46.0'
                elem = elem.strip()
                new_row.append(elem)
            cleaned.append(new_row)
        return np.array(cleaned)

    X_str = clean_cell_array(X_raw)
    y_str = clean_cell_array(y_raw).reshape(-1)

    # Step 2: convert X to float
    X = X_str.astype(float)

    # Step 3: convert y to numeric or keep string
    try:
        y = y_str.astype(float)
    except:
        y = y_str.astype(str)

    return X, y