A Data-Driven Artificial Neural Network Approach to Software Project Risk Assessment

Algorithm 1

Proposed ANN model.

1. Load the Dataset

2. Pre-process to omit and sort NULL values

3. Split the dataset for training and testing (80% for training and 20% for testing)

4. Visualize the data for each factor in relation to other factors

5. Train the dataset

6. Use the designed ANN model with five input values, 1 hidden dense layer and 1 output value

7. Use MSE/MAE with ANN Model

8. Run cycles and Store the history

9. Plot the error as loss function

10. Check Robustness of the model through Diebold–Mariano test

11. After minimal loss, forecast the total risk for next project.