Artificial Neural Network Project

Project Overview

This project, titled "KTU LAB4_2", involves building and training an artificial neural network (ANN) to predict loan statuses based on a given dataset. The project is implemented using Jupyter Notebook and primarily uses TensorFlow and Keras libraries.

Dataset

The dataset used in this project is credit_risk_dataset.csv, which contains information about individuals' loan applications and statuses. The dataset includes the following columns:

• person_age
• person_income
• person_home_ownership
• person_emp_length
• loan_intent
• loan_grade
• loan_amnt
• loan_int_rate
• loan_status
• loan_percent_income
• cb_person_default_on_file
• cb_person_cred_hist_length

Data Preprocessing

The data is first cleaned by handling missing values and converting categorical variables into dummy/indicator variables. The dataset is then split into training and testing sets.

Model Architecture

The ANN model consists of:

• An input layer with 10 neurons and ReLU activation
• Two hidden layers with 5 neurons each and ReLU activation
• An output layer with a single neuron and a linear activation function

model = Sequential()
model.add(Dense(10, input_dim=X_train.shape[1], activation='relu'))
model.add(Dense(5, activation='relu'))
model.add(Dense(1, activation='linear'))

Training the Model

The model is compiled using the mean squared error loss function and the Adam optimizer. It is trained for 20 epochs with a batch size of 64.

model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse'])
history = model.fit(X_train, y_train, epochs=20, batch_size=64, validation_split=0.2)

Evaluation

The model's performance is evaluated using the mean squared error (MSE) metric. Additionally, cross-validation is performed to estimate the model's accuracy.

Analysis Using Project Pictures

Activation Functions and Their Derivatives

(#)

XOR with Different Activation Functions

(#)

Training History

Cross-Validation Results

Improvements

To increase the model's accuracy, the following changes were made:

• Added additional hidden layers
• Adjusted the learning rate
• Changed the activation function of the output layer to sigmoid

def create_improved_model():
    model = Sequential()
    model.add(Dense(15, input_dim=X_train.shape[1], activation='relu'))
    model.add(Dense(10, activation='relu'))
    model.add(Dense(5, activation='relu'))
    model.add(Dense(1, activation='sigmoid'))
    optimizer = Adam(learning_rate=0.001)
    model.compile(loss='binary_crossentropy', optimizer=optimizer, metrics=['accuracy'])
    return model

Results

The improved model achieved an average accuracy of 0.79, which is a 4.44% improvement over the initial model. However, this did not meet the target of a 5% improvement.

Conclusion

This project demonstrates the process of building, training, and evaluating an artificial neural network for loan status prediction. Despite improvements, further optimization is required to achieve the desired accuracy.

Feel free to explore the notebook for more details.

You can add this content by navigating to the repository and creating a new file named `README.md`.