README.md

---
tags:
  - random-forest
  - regression
  - land-surface-temperature
  - remote-sensing
  - landsat
library_name: sklearn
---

# LST Random Forest Regression Model

Predicts **Land Surface Temperature (°C)** from Landsat 9 spectral indices.

## Features
- `longitude`
- `latitude`
- `ndvi`
- `ndbi`
- `ndwi`
- `elevation`
- `albedo`

## Target
- `lst_c` — Land Surface Temperature in Celsius

## Performance
| Split | RMSE | MAE | R² |
|-------|------|-----|----|
| Train | 0.7878 | 0.5133 | 0.9891 |
| Test  | 2.0728 | 1.4003 | 0.9288 |

5-fold CV R²: **0.9178 ± 0.0183**

## Usage
```python
import joblib, numpy as np
rf = joblib.load("rf_lst_model.joblib")
# [longitude, latitude, ndvi, ndbi, ndwi, elevation, albedo]
sample = np.array([[121.7, 31.2, 0.33, -0.09, -0.40, 4.0, 0.18]])
print(rf.predict(sample))
```