README.md

metadata

title: WaterWise Home
emoji: 💧
sdk: docker
app_port: 7860
pinned: false
license: mit

WaterWise Home

Your smart water assistant. Log your daily water-using activities, choose your season, and instantly get 3 personalized tips to help you save more. Tracks your lifetime savings and keeps a local history for demos.

Why We Built It

Household water waste adds up — long showers, half-full laundry loads, watering plants at noon. WaterWise Home gives simple, personalized guidance that’s easy and instant.

✨ Features

• AI-powered recommendations (MiniLM + cosine, top-3)
• Season-aware tips (Winter / Spring / Summer / Fall)
• Lifetime Water Saved counter + local demo history
• Clean, responsive UI with one-click examples

How It Works

1. You describe an activity (e.g., “Ran a full dishwasher”).
2. (Optional) Choose your season.
3. The app embeds your text, searches 1,000+ tips, and returns the top-3 most similar.

Dataset (Required by brief)

• Public HF dataset: https://huggingface.co/datasets/Liat2025/waterwise_tips_2025
• Loaded directly in the app from the CSV URL (no local file).
• Size: 1,000+ tips • Column: tip (UTF-8, one tip per row)

Mini Evaluation

Query: “Watered the garden at noon (summer)”
Engine: sentence-transformers/all-MiniLM-L6-v2 + cosine (top-3)

Example output:

1. Water in early morning or evening to reduce evaporation
2. Use drip irrigation / soaker hose instead of sprinklers
3. Add mulch around plants to retain soil moisture

Tech Stack

Gradio, Python, sentence-transformers, torch, pandas, numpy.

Run Locally

git clone https://huggingface.co/spaces/Liat2025/WaterWiseHomecopy
cd WaterWiseHomecopy
pip install -r requirements.txt
python app.py