Creating classification models and evaluating accuracy of spam detection

Uses previous student loan recipients to develope a model to predict the likelihood that an applicant will repay their student loans. Applies understanding of neural networks to develop a more accurate load repayment prediction model and more accurate interest rates for borrowers

The primary skills demonstrated include the ability to:

• Prepare data for neural network model use
• Compile and evaluate a prediction model using neural network
• Create, train, and evaluate neural network models using TensorFlow
• Use KerasTuner to assist with finding optimal neural network structures
• Save trained TensorFlow models for later use

Python Libraries and Functions

• pandas
• pathlib
• tensorflow
• Dense (from tensorflow.keras.layers)
• Sequential (from tensorflow.keras.models)
• train_test_split (from klearn.model_selection)
• classification_report (from sklearn.metrics)
• StandardScaler (from sklearn.preprocessing)

Jamie Bond | [email protected] | Connect on Linkedin

Adapted from instructions starter files provided by The Artificial Intelligence Boot Camp at UNC Charlotte, including:

• Data: https://static.bc-edx.com/ai/ail-v-1-0/m18/lms/datasets/student-loans.csv
• Instructions: student_loans_with_deep_learning.ipynb