House Price Predictor
Budget: ₹600 – ₹1,500 INR
I’ve collected a sizable housing-market dataset and now need a complete prediction pipeline built in Python. The work starts with thorough cleaning, preprocessing, and creative feature engineering so that every useful signal—size, rooms, location ratings, and any other attributes we can derive—is captured.
Once the dataset is ready, I want to benchmark several approaches, but the core focus is Random Forest Regression. Feel free to test Linear or Support Vector methods if they might edge out the forest, yet the final report should clearly show how each model performs against the usual metrics (RMSE, MAE, R²).
Visual insight is important, so the notebook or script has to generate intuitive plots—scatter trends, residual diagnostics, whatever best tells the story of model quality and feature impact. After modelling, package the chosen estimator in a lightweight Streamlit app where a user can key in property details and instantly see a predicted price.
Deliverables
• Clean, well-commented Python code (Pandas, NumPy, scikit-learn, Matplotlib / Seaborn, Streamlit)
• A concise report explaining preprocessing steps, feature choices, model comparison results, and recommendations
• The Streamlit interface ready to run locally (with instructions)
• Any supplementary documentation needed to reproduce the full workflow end to end
If something in the data suggests additional techniques—hyper-parameter tuning, cross-validation strategies, or advanced visualisations—please flag it and incorporate where it boosts accuracy or interpretability. I look forward to your expertise turning raw rows into a reliable, user-friendly pricing tool.
Once the dataset is ready, I want to benchmark several approaches, but the core focus is Random Forest Regression. Feel free to test Linear or Support Vector methods if they might edge out the forest, yet the final report should clearly show how each model performs against the usual metrics (RMSE, MAE, R²).
Visual insight is important, so the notebook or script has to generate intuitive plots—scatter trends, residual diagnostics, whatever best tells the story of model quality and feature impact. After modelling, package the chosen estimator in a lightweight Streamlit app where a user can key in property details and instantly see a predicted price.
Deliverables
• Clean, well-commented Python code (Pandas, NumPy, scikit-learn, Matplotlib / Seaborn, Streamlit)
• A concise report explaining preprocessing steps, feature choices, model comparison results, and recommendations
• The Streamlit interface ready to run locally (with instructions)
• Any supplementary documentation needed to reproduce the full workflow end to end
If something in the data suggests additional techniques—hyper-parameter tuning, cross-validation strategies, or advanced visualisations—please flag it and incorporate where it boosts accuracy or interpretability. I look forward to your expertise turning raw rows into a reliable, user-friendly pricing tool.