🌦️ Weather Temperature Prediction using RNN (Deep Learning)

A deep learning project focused on building and evaluating a Recurrent Neural Network (RNN) for time-series forecasting using weather data.

---

🚀 Run Notebook in Google Colab

---

📘 Project Overview

This project implements a SimpleRNN model to predict future temperature values based on historical weather data.

The project covers:

• Dataset exploration

• Time-series preprocessing

• Sequence generation

• RNN model design

• Model training and validation

• Model evaluation

• Forecast visualization

---

🎯 Objective

• Understand time-series dataset structure

• Perform data preprocessing and normalization

• Create sequences for RNN input

• Build a SimpleRNN model using TensorFlow/Keras

• Evaluate model performance

• Forecast future temperature values

---

📂 Dataset Information

Dataset: Daily Weather Dataset (Kaggle)

https://www.kaggle.com/datasets/muthuj7/weather-dataset

Feature	Description
Date	Timestamp
Temperature	Target variable
Humidity	Atmospheric moisture
Wind Speed	Wind velocity
Pressure	Atmospheric pressure

---

🧹 Data Preprocessing

• Data exploration

• Missing value handling

• Feature scaling using MinMaxScaler

• Sequence creation (7–14 days input)

• Train / Validation / Test split

---

🧠 RNN Model Architecture

• Input Layer

• SimpleRNN Layer (32–64 units)

• Dropout Layer

• Dense Output Layer

---

⚙ Model Training

• Batch training

• Validation monitoring

• Loss curve analysis

---

📊 Model Evaluation

• Mean Absolute Error (MAE)

• Root Mean Squared Error (RMSE)

• R² Score

---

🔍 Prediction Visualization

• Actual vs Predicted comparison

• Trend visualization

• Forecast accuracy evaluation

---

🔮 Forecasting

• Predict next 7 days temperature

• Uses recent historical data

---

⚠️ Limitations

• SimpleRNN struggles with long-term dependencies compared to LSTM/GRU

• Limited dataset size may affect generalization

• External weather factors are not included

• Performance may decrease for long-term forecasting

• Model depends heavily on data quality

---

🛠 Tech Stack

Tool	Purpose
Python	Programming
TensorFlow / Keras	Deep Learning
NumPy	Computation
Pandas	Data handling
Matplotlib	Visualization
Scikit-learn	Preprocessing
Kaggle API	Dataset download
Google Colab	Development

---

📁 Repository Structure

weather-rnn-temperature-prediction/

│
├── DL_Assignment_2_RNN.ipynb

├── README.md

└── DL Assignment 2 - RNN.pdf

---

🚀 How to Run the Project

1️⃣ Install Required Libraries

pip install tensorflow pandas numpy matplotlib scikit-learn kaggle

---

2️⃣ Kaggle API Setup & Dataset Download

mkdir -p ~/.kaggle
cp kaggle.json ~/.kaggle/
chmod 600 ~/.kaggle/kaggle.json

kaggle datasets download -d muthuj7/weather-dataset
unzip weather-dataset.zip

---

3️⃣ Run the Notebook

• Open notebook in Google Colab

• Run all cells sequentially

• View predictions and visualizations

---

📌 Academic Submission

This project was created as part of a Deep Learning assignment, demonstrating the implementation of a time-series forecasting model using Recurrent Neural Networks (RNN), including data preprocessing, sequence generation, model training, evaluation, and prediction visualization.

---

📌 Future Enhancements

• Implement advanced models like LSTM and GRU

• Perform hyperparameter tuning

• Use larger and more diverse datasets

• Integrate real-time weather APIs

• Add interactive dashboards

---

👤 Author

Name: Laya Mary Joy

Organization: Entri Elevate

Date: March 21, 2026

---

⭐ Acknowledgment

Thanks to Entri Elevate for guidance and support.

---