Numerical Data Classification Model
Budget: ₹12,500 – ₹37,500 INR
I have a structured, purely numerical dataset and need a robust classification model built from it. Your job is to take the raw data, carry out the full machine-learning workflow, and hand back a clean, ready-to-deploy solution.
Scope of work
• Inspect and clean the dataset, flagging any outliers or missing values.
• Explore key correlations and feature importance to justify modeling choices.
• Build and compare at least two classification algorithms (e.g., random forest, XGBoost, or any other well-suited approach).
• Tune hyperparameters for best accuracy while guarding against overfitting.
• Deliver concise performance metrics—confusion matrix, precision, recall, F1, ROC-AUC—plus a brief narrative on why the chosen model is optimal.
• Package reproducible code (Python, scikit-learn or similar) with clear comments and a short README so I can rerun everything end-to-end.
I’m looking for clean, understandable code, well-explained reasoning, and a model that generalizes. If you have prior examples of numerical-data classification projects, feel free to reference them; otherwise, a quick outline of your proposed workflow will help me decide quickly.
Scope of work
• Inspect and clean the dataset, flagging any outliers or missing values.
• Explore key correlations and feature importance to justify modeling choices.
• Build and compare at least two classification algorithms (e.g., random forest, XGBoost, or any other well-suited approach).
• Tune hyperparameters for best accuracy while guarding against overfitting.
• Deliver concise performance metrics—confusion matrix, precision, recall, F1, ROC-AUC—plus a brief narrative on why the chosen model is optimal.
• Package reproducible code (Python, scikit-learn or similar) with clear comments and a short README so I can rerun everything end-to-end.
I’m looking for clean, understandable code, well-explained reasoning, and a model that generalizes. If you have prior examples of numerical-data classification projects, feel free to reference them; otherwise, a quick outline of your proposed workflow will help me decide quickly.