satwikshreshth1/smart-env-monitor-models

Smart Environment Monitor — ML Models

These are the trained ML models for the smart-env-monitor project — a Raspberry Pi based indoor environment monitoring system that measures temperature, humidity, and air quality (CO₂) and predicts air quality labels in real-time.

---

Models

File	Type	Purpose
models/rf_model.pkl	Random Forest	Predicts air quality label (Good / Moderate / Poor / Hazardous)
models/kmeans_model.pkl	K-Means (k=4)	Clusters readings into 4 groups for borderline detection
models/isoforest_model.pkl	Isolation Forest	Detects anomalous sensor readings
models/label_encoder.pkl	Label Encoder	Encodes/decodes air quality labels
models/scaler.pkl	Standard Scaler	Scales features for K-Means input

---

Input Features

Feature	Description
temp_c	Temperature in °C
humidity	Relative humidity in %
co2_ppm	CO₂ in ppm (derived from MQ135 raw ADC reading)

---

Labels

Label	Meaning
Good	Clean air
Moderate	Acceptable air quality
Poor	Ventilation recommended
Hazardous	Immediate action required

---

K-Means Cluster Mapping

Cluster	Label
0	Good
1	Poor
2	Hazardous
3	Moderate

---

Usage

from huggingface_hub import hf_hub_download
import joblib
import os

REPO = "satwikshreshth1/smart-env-monitor-models"
FILES = [
    "models/rf_model.pkl",
    "models/kmeans_model.pkl",
    "models/isoforest_model.pkl",
    "models/label_encoder.pkl",
    "models/scaler.pkl",
]

os.makedirs("models", exist_ok=True)
for f in FILES:
    hf_hub_download(repo_id=REPO, filename=f, local_dir=".")

rf     = joblib.load("models/rf_model.pkl")
scaler = joblib.load("models/scaler.pkl")

---

Hardware

• Microcontroller: Arduino (serial output)
• Sensors: DHT11/DHT22 (temp & humidity), MQ135 (air quality)
• Host: Raspberry Pi (runs inference)

---

License

MIT