draft.txt

- Fit a logistic regression model on the training data to predict the loan status
  - I have a set of feature vectors (set of horizontal vectors)
  - I have a target vector (vertical) (Loan_Status)
  - I have an initial parameters (weights) vector (horizontal)

wj_improved = wj - α*sum((sigmoid(parameters_vector . feature_vectors[i]) - target_vector[i]) * feature_vectors[i][j])

Steps
  - Pick a learning rate (the training data contains about 614 rows. So pick a learning rate and I'll tune it)
  - Generate an initial parameters (weights) vector whose length is the length of a features vector
    (len(processed_features_train.columns) or something like that)
  - for j in range len(parameters_vector):
    - parameters_vector[j] -=
        learning_rate * sum(
          (1/(1 + e ** (-parameters_vector . feature_vectors[i])) - target_vector[i]) * feature_vectors[i][j]
          for i in range len(feature_vectors)
        )

chromium print options: portrait, A1, scale:175, no headers/footers, background graphics