Loan Prediction Model with Imbalanced Data Handling

Cliente Freelancer · Remoto · Remoto · freelance · mid · 12.500–37.500 INR

Publicada el 2026-07-19

Descripción de la oferta

I have a historical lending dataset that requires a full end-to-end machine-learning workflow. First, I need thorough data preprocessing and exploratory analysis: identify and treat missing values, detect and cap or remove outliers, and inspect the class distribution. In this project the key concern is minority class under-representation, so any resampling or cost-sensitive approach you propose should directly address that imbalance. For feature engineering I want, at minimum, robust encoding of all categorical variables and consistent scaling or normalizing of the numeric fields before they enter the models. Feel free to add creative interactions if they genuinely improve recall without overfitting. Once the data pipeline is solid, train and compare several classification algorithms—Logistic Regression, Random Forest, XGBoost and, if you believe it will help, a simple neural network. Optimisation focuses on accuracy, precision, recall and ROC-AUC, but recall carries the highest weight because false negatives (missed defaulters) are the most expensive error for me. Deliverables: • Cleaned, feature-engineered dataset ready for inference • Jupyter or Colab notebook (or well-documented Python scripts) that reproduces preprocessing, modelling, and evaluation • Comparative metrics table highlighting the best performing model by recall and ROC-AUC • Saved model file plus load-and-predict script Bonus (preferred): wrap the chosen model in a lightweight Streamlit or Flask web app so I can upload a CSV row or enter values manually and see the prediction along with the key probability score. Please outline your proposed methodology, libraries (I mainly use pandas, scikit-learn, imbalanced-learn, XGBoost, TensorFlow/Keras) and any additional steps you find valuable. I am ready to start as soon as I select the right collaborator.

Skills

• Machine Learning
• Pandas
• Python
• Microsoft Excel
• Data Science
• Power BI
• Business Intelligence
• scikit-learn
• R
• TensorFlow
• CAP
• Jupyter
• Flask
• Keras

Fuente original: freelancer