New method to load data and new release 0.0.23

[gaetanbrison]

Hi Jun 👋

Adding a new method to load all datasets of benchmarks from Hugging Face as follow:

# Import the required dataset loading module from Hugging Face
from datasets import load_dataset

# Load the dataset from the Hugging Face Hub, specifying the source file and the split
dataset = load_dataset('inria-soda/carte-benchmark', data_files='data_raw/michelin.csv', split='train')

# Convert the loaded dataset to a Pandas DataFrame for easier manipulation
df = dataset.to_pandas()

# Import the LabelEncoder from scikit-learn to encode categorical labels
from sklearn.preprocessing import LabelEncoder
encoder = LabelEncoder()

# Encode the target column ("Award") to numerical values for machine learning tasks
df["Award"] = encoder.fit_transform(df["Award"])

# Define the number of training samples/entities and set a random state for reproducibility
num_train = 128  # Number of training groups or entities
random_state = 1  # Random seed for reproducibility
target_name = "Award"  # Target column name
entity_name = "Name"  # Entity column name

# Split the dataset into training and testing sets using a custom function, setting a random seed
X_train, X_test, y_train, y_test = set_split_hf(df, target_name, entity_name, num_train, random_state=42)