Contact person: Taxi Fare Prediction Web App

Phone:Show

Email:Show

Location: Cairo, Egypt

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"I have an initial taxi-fare dataset in CSV form and need a complete end-to-end solution that lets anyone enter trip details and instantly see the predicted fare. The core data is ready, but I expect you may want to enrich it with additional public taxi records to boost model performance.

You’ll start with thorough cleaning, feature engineering (distance calculations, temporal splits, categorical handling, etc.), then train and tune a regression model—XGBoost is my first choice, though I’m open to seeing comparative results from alternatives if you believe they add value. Model quality should be demonstrated with R², MSE, and any other metrics that highlight over-/underfitting or bias.

Once we’re happy with the metrics, wrap the predictor in a lightweight web UI built in either Gradio or Streamlit. The interface must allow users to supply pickup/drop-off coordinates, timestamps, passenger count, and instantly receive a fare estimate.

Finally, deploy the app publicly. I have no fixed preference between Hugging Face Spaces, Google Cloud, or another reliable host, so choose whichever lets us launch quickly while keeping costs sensible and maintenance minimal.

Deliverables:
• Well-commented Python code and a Jupyter notebook that walks through data prep, training, evaluation, and model export
• Requirements file or environment spec for reproducibility
• Web app source integrated with the trained model
• Working live deployment URL and concise deployment instructions
• Brief documentation explaining data sources, feature choices, hyper-parameter decisions, and future improvement ideas

Clear, readable code and step-by-step explanations are more important to me than flashy visuals, so please prioritise maintainability and clarity throughout." (client-provided description)