config.yaml

# --- Paths ---
paths:
  data:
    dataset: "machine_learning/data/raw/dataset.json"        # Path to the entire raw dataset
    train_set: "machine_learning/data/processed/train.csv"       # Path to save the training dataset
    test_set: "machine_learning/data/processed/test.csv"         # Path to save the test dataset

  models:
    logistic_regression: "machine_learning/models/logistic_regression.pkl"
    naive_bayes: "machine_learning/models/naive_bayes.pkl"
    svm: "machine_learning/models/svm.pkl"
    tfidf_vectorizer: "machine_learning/models/tfidf_vectorizer.pkl"

  results:
    model_results: "machine_learning/results/model_results.csv"


# --- Training Parameters ---
# --- Training Parameters ---
training:
  test_size: 0.2                      # Proportion of the dataset to include in the test split (20%)
  random_state: 42                    # Random seed for reproducibility of train-test splits
  tfidf_max_features: 5000            # Maximum number of features (unique words or n-grams) to retain in TF-IDF
  tfidf_ngram_range: [1, 2]           # N-gram range for TF-IDF: [1, 2] includes unigrams and bigrams
  logistic_regression_max_iter: 1000  # Maximum number of iterations for Logistic Regression's solver to converge
  naive_bayes_alpha: 0.5              # Additive (Laplace/Lidstone) smoothing parameter for Naive Bayes
  svm_kernel: "linear"                # Kernel type for SVM: 'linear', 'rbf', 'poly', etc., determine decision boundaries
  svm_c: 1.0                          # Regularization parameter for SVM: smaller values for stronger regularization



# --- API Parameters (when applicable later) ---
# api:
#   host: "127.0.0.1"
#   port: 5000