Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data

KO | EN

gitlite — search

#python #react #ai #typescript #android #machine-learning #linux #macos #awesome #cli #kubernetes #security #ios #open-source

Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data

(★ 15)

This project applies machine learning to predict crop yield using synthetic remote sensing and weather data, showcasing how vegetation indices and climatic factors can influence agricultural productivity. It demonstrates yield forecasting with Random Forest and visual insights into feature importance.

Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data Latest Version Download

Download Latest Version (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# 🌾 Crop Yield Prediction Using Remote Sensing & Weather Data

![Python](https://img.shields.io/badge/Python-3.8+-3d6b4a?style=for-the-badge&logo=python&logoColor=white)
![scikit-learn](https://img.shields.io/badge/scikit--learn-1.4+-e8b03a?style=for-the-badge&logo=scikit-learn&logoColor=white)
![Streamlit](https://img.shields.io/badge/Streamlit-Dashboard-c97a20?style=for-the-badge&logo=streamlit&logoColor=white)
![License](https://img.shields.io/badge/License-MIT-3d6b4a?style=for-the-badge)
![Status](https://img.shields.io/badge/Status-Active-brightgreen?style=for-the-badge)

**An end-to-end ML pipeline fusing satellite-derived vegetation indices with climate variables to forecast crop productivity — scalable to real-world Sentinel-2 & Landsat datasets.**

[🚀 Run Dashboard](https://crop-yield-prediction-using-remote-sensing-weather-data-jcs2k6.streamlit.app/) · [📊 View Results](#-model-results--visualizations) · [📐 Methodology](#-methodology) · [👤 Author](#-author)

</div>

---

## 📌 Project Overview

This project builds an integrated agricultural analytics pipeline that combines **vegetation health indices** (NDVI, EVI), **weather variables** (Rainfall, Temperature, Humidity), and **soil data** (Soil Moisture) to predict crop yield in tons per hectare (t/ha).

A **Random Forest Regressor** is used for its ability to capture non-linear environmental interactions and provide interpretable feature importance rankings. The framework is designed to scale seamlessly with real satellite datasets such as Sentinel-2 and Landsat.

### 🎯 Objectives

- Build an integrated agricultural dataset combining vegetation and climate indicators
- Develop a reproducible ML workflow for yield forecasting
- Evaluate model performance using robust regression metrics
- Interpret environmental drivers influencing yield
- Deliver a portfolio-ready environmental ML project

### 🌍 Applications

| Domain | Use Case |
|---|---|
| 🌱 Precision Agriculture | Targeted resource allocation based on predicted yield zones |
| 🌍 Climate-Smart Farming | Adaptive decisions under changing climate conditions |
| 📊 Food Security Planning | Data-driven national and regional food policy insights |
| 🛰 Remote Sensing Ops | Scalable to Sentinel-2, Landsat, and GIS-ready workflows |

---

## 📊 Key Results at a Glance

| Metric | Value | Description |
|:---:|:---:|:---|
| 🎯 **R² Score** | **0.855** | 85.5% of yield variance explained |
| 📉 **RMSE** | **1.026 t/ha** | Average prediction error |
| 🗃 **Dataset** | **200 samples** | 6 environmental features |
| 🌾 **Mean Yield** | **6.67 t/ha** | Range: −2.3 to 12.2 t/ha |

</div>

---

## 🗂 Project Structure

```
Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data/
│
├── 📊 crop_yield_data.xlsx        # Synthetic dataset (200 samples, 6 features)
├── 🐍 crop_yield_prediction.py    # Core ML training & visualization script
├── 🖥  dashboard.py               # Streamlit interactive dashboard
├── 📋 requirements.txt            # Project dependencies
└── 📖 README.md                   # Documentation
```

---

## 📋 Dataset & Features

**Type:** Synthetic · **Samples:** 200 · **Purpose:** ML workflow demonstration

| Feature | Category | Description | Range | Importance |
|---|:---:|---|:---:|:---:|
| 🌡 **Temperature (°C)** | Weather | Surface air temperature | 15–40 °C | 🟩🟩🟩🟩🟩 53.6% |
| 🌧 **Rainfall (mm)** | Weather | Total precipitation | 100–1200 mm | 🟩🟩🟩 26.3% |
| 🛰 **NDVI** | Remote Sensing | Normalized Difference Vegetation Index | 0.20–0.89 | 🟨 8.8% |
| 💧 **Soil Moisture** | Soil | Root-zone water availability | 0.10–0.50 | 🟧 5.7% |
| 💦 **Humidity (%)** | Weather | Atmospheric moisture content | 30–90 % | ⬜ 3.0% |
| 🌿 **EVI** | Remote Sensing | Enhanced Vegetation Index | 0.10–0.79 | ⬜ 2.7% |
| **🌾 Crop Yield (t/ha)** | 🎯 **Target** | Crop productivity in tons per hectare | −2.3–12.2 | — |

---

## 📐 Methodology

```
┌─────────────────────────────────────────────────────────────┐
│                    ML PIPELINE WORKFLOW                     │
├──────────┬──────────┬──────────┬──────────────────────────┤
│  STEP 1  │  STEP 2  │  STEP 3  │         STEP 4           │
│    📥    │    🌲    │    📐    │           🔍             │
│  Data    │  Model   │  Evalu-  │      Interpretation      │
│  Prep    │ Training │  ation   │                          │
├──────────┼──────────┼──────────┼──────────────────────────┤
│ Load     │ Random   │ RMSE     │ Feature importance       │
│ Excel    │ Forest   │ & R²     │ ranking                  │
│ dataset  │ 100 trees│ scores   │                          │
│          │          │          │                          │
│ 80/20    │ Captures │ Scatter  │ Environmental yield      │
│ split    │ nonlinear│ plots    │ driver analysis          │
│          │ patterns │          │                          │
└──────────┴──────────┴──────────┴──────────────────────────┘
```

### 🌲 Why Random Forest?

- Handles **nonlinear** environmental relationships naturally
- **Robust** on tabular agricultural data
- Reduces overfitting through **ensemble learning**
- Provides built-in **feature importance** for interpretability

---

## 📊 Model Results & Visualizations

### 1️⃣ Feature Importance

```
Temperature (°C)  ████████████████████████████████████████  53.6%
Rainfall (mm)     ████████████████████                       26.3%
NDVI              ██████                                      8.8%
Soil Moisture     ████                                        5.7%
Humidity (%)      ██                                          3.0%
EVI               ██                                          2.7%
```

> **Key insight:** Temperature is the dominant predictor, followed by Rainfall — confirming that thermal and water stress are primary drivers of yield in this dataset.

---

### 2️⃣ Feature Correlation with Crop Yield

```
Rainfall (mm)     ████████████████  +0.416  (positive — more rain → higher yield)
NDVI              ██████████        +0.325  (positive — healthier vegetation → higher yield)
Soil Moisture     ████              +0.150  (positive — wetter soil → higher yield)
Humidity (%)      ▏                 +0.012  (negligible)
EVI               ████████         −0.122  (weakly negative)
Temperature (°C)  █████████████    −0.430  (negative — heat stress → lower yield)
```

---

### 3️⃣ Actual vs Predicted — Test Set

```
Yield   12 |        ×
(t/ha)  10 |  ·  ×      ×         ×    ×
         8 |    ·   ×       ×  ×     ×
         6 |  ×   ·    ·  ×       ·    ×
         4 |    ×    ·   ×    ·       ×
         2 |  ×    ·                ×
         0 | ×
        -2 |__________________________________
            0    5   10   15   20   25   30   40
                        Sample Index
                 · Actual   × Predicted
```

> Model tracks actual values closely across the full yield range. Larger deviations occur at yield extremes (very low or very high), a known limitation of ensemble methods on sparse edge cases.

---

### 4️⃣ Yield Distribution — All 200 Samples

```
Count
 31 |                          ████
 29 |                     ████ ████
 26 |                     ████ ████ ████
 22 |               ████  ████ ████ ████
 21 |          ████ ████  ████ ████ ████
 15 |          ████ ████  ████ ████ ████ ████
 14 |     ████ ████ ████  ████ ████ ████ ████
  9 | ████ ████ ████ ████ ████ ████ ████ ████ ████
  1 |─────────────────────────────────────────────────
     -2.3 -0.4  1.6  3.5  5.4  7.4  9.3 11.2
                      Yield (t/ha)
```

> Distribution is approximately normal, centred around 6–8 t/ha, with a slight left tail at negative yields (synthetic data artefact).

---

### 5️⃣ NDVI vs Crop Yield (Scatter)

```
Yield  12 |                    ·   ·
(t/ha) 10 |          ·   ·  ·   ·     · ·
        8 |    ·  ·     ·   ·  · ·  ·  ·  ·
        6 |  ·   ·  ·  ·  ·   ·    ·  ·  ·
        4 |  · ·   ·     ·  ·    ·  ·
        2 |    · ·   ·          ·
        0 |  ·
       -2 |__________________________________________
           0.20  0.30  0.40  0.50  0.60  0.70  0.80  0.90
                               NDVI
```

> A moderate positive trend (r = +0.325) — healthier vegetation (higher NDVI) generally correlates with higher yield, though Temperature and Rainfall remain stronger predictors.

---

## 💡 Key Insights

1. **Temperature is the primary driver** — accounting for 53.6% of feature importance; heat stress significantly suppresses yield
2. **Rainfall has strong secondary influence** — 26.3% importance and r = +0.416 with yield
3. **NDVI correlates positively** with yield (r = +0.325), confirming satellite vegetation monitoring is a valid yield proxy
4. **Ensemble learning outperforms** single-tree models by reducing variance across heterogeneous environmental data
5. **Integrating remote sensing with weather data** creates a more stable and interpretable prediction framework than either data source alone

---

## ▶️ How to Run

### 1. Clone the Repository

```bash
git clone https://github.com/Nelvinebi/Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data.git
cd Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data
```

### 2. Install Dependencies

```bash
pip install -r requirements.txt
```

### 3a. Run the ML Script

```bash
python crop_yield_prediction.py
```

### 3b. Launch the Streamlit Dashboard

```bash
streamlit run dashboard.py
```

> ⚠️ Ensure `crop_yield_data.xlsx` is in the same directory as the script.

---

## 🛠 Tools & Technologies

| Tool | Version | Purpose |
|---|:---:|---|
| ![Python](https://img.shields.io/badge/-Python-3d6b4a?logo=python&logoColor=white) | 3.8+ | Core language |
| ![scikit-learn](https://img.shields.io/badge/-scikit--learn-e8b03a?logo=scikit-learn&logoColor=white) | 1.4+ | ML modelling & evaluation |
| ![Pandas](https://img.shields.io/badge/-Pandas-173350?logo=pandas&logoColor=white) | 2.0+ | Data manipulation |
| ![NumPy](https://img.shields.io/badge/-NumPy-3d6b4a?logo=numpy&logoColor=white) | 1.26+ | Numerical computing |
| ![Matplotlib](https://img.shields.io/badge/-Matplotlib-c97a20?logoColor=white) | 3.8+ | Static visualization |
| ![Plotly](https://img.shields.io/badge/-Plotly-255080?logo=plotly&logoColor=white) | 5.20+ | Interactive charts |
| ![Streamlit](https://img.shields.io/badge/-Streamlit-c97a20?logo=streamlit&logoColor=white) | 1.32+ | Web dashboard |
| ![OpenPyXL](https://img.shields.io/badge/-OpenPyXL-6fa87e?logoColor=white) | 3.1+ | Excel file I/O |

---

## ⚠️ Limitations

- Dataset is **synthetic** — not derived from real satellite imagery
- No **temporal sequence modelling** (crops are inherently seasonal)
- No **spatial GIS integration** or coordinate data
- Minimal **hyperparameter tuning** performed
- Single crop type — no multi-crop generalisation

---

## 🔮 Future Improvements

- [ ] Integrate real **Sentinel-2 / Landsat** satellite datasets
- [ ] Add **time-series modelling** with LSTM or Temporal CNN
- [ ] Perform **hyperparameter optimisation** via GridSearchCV / Optuna
- [ ] Deploy interactive **Streamlit / Flask web dashboard**
- [ ] GIS-based **spatial yield mapping** with GeoPandas & Folium
- [ ] **Multi-crop** dataset expansion and generalisation

---

## 👤 Author

**AGBOZU EBINGIYE NELVIN**

*Environmental Data Scientist · GIS · Remote Sensing · Machine Learning*

End-to-end environmental intelligence solutions for flood risk, water quality, vegetation monitoring, and climate-resilient planning.

[![GitHub](https://img.shields.io/badge/GitHub-Nelvinebi-3d6b4a?style=for-the-badge&logo=github)](https://github.com/Nelvinebi)
[![LinkedIn](https://img.shields.io/badge/LinkedIn-agbozu--ebi-255080?style=for-the-badge&logo=linkedin)](https://www.linkedin.com/in/agbozu-ebi/)

</div>

---

🌾 *"Bridging satellite intelligence and machine learning for a food-secure future."*

</div>

Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data

Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data

Crop-Yield-Prediction-Using-Remote-Sensing-Weather-Data Latest Version Download

Today's picks

Was this content helpful?

Recent Feedback

Download README