Weather Trend Forecasting Project

Project Name: Weather Trend Forecasting (Basic)
Part of: Company PM Accelerator Program
Objective: Predict next-day temperature (°C) using historical weather data with Ridge Regression.

---

Table of Contents

• Project Overview
• Key Features
• Installation
• Dataset
• Usage
• Model Details
• Results
• Limitations
• Future Improvements
• Contributing
• License
• Contact

---

Project Overview

This project demonstrates a basic implementation of weather trend forecasting using Ridge Regression. The goal is to predict the next day's temperature (°C) based on historical weather data, including features like wind speed, precipitation, humidity, and pressure. The code is designed for the Company PM Accelerator Program to showcase data cleaning, exploratory analysis, and regression modeling.

---

Key Features

• Data Cleaning: Filters 12 core weather features from raw data.
• Exploratory Analysis: Visualizes temperature and precipitation trends.
• Time-Shifted Target: Creates a target variable for next-day temperature prediction.
• Ridge Regression Model: Trains and evaluates a linear model with regularization.
• Model Evaluation: Reports Mean Absolute Error (MAE) and visualizes predictions vs. actuals.

---

Installation

1. Clone the Repository:

   git clone https://github.com/your-company/pm-accelerator-weather-forecasting.git
   cd pm-accelerator-weather-forecasting

Install Dependencies: pandas==1.3.5 scikit-learn==1.0.2 matplotlib==3.5.1 2. Dataset Source: GlobalWeatherRepository.csv (not included in this repository for licensing reasons). Required Columns: latitude (used as index in the code) temperature_celsius, temperature_fahrenheit wind_mph, wind_kph, wind_degree pressure_mb, pressure_in precip_mm, precip_in humidity, visibility_km, visibility_miles

Place the dataset in the root directory before running the code

3. Usage

   1. Data Preparation Ensure the dataset is named GlobalWeatherRepository.csv and placed in the project root. Run the Jupyter notebook or Python script to load and preprocess data: weather = pd.read_csv("GlobalWeatherRepository.csv", index_col="latitude")
   2. Exploratory Analysis Generate plots for temperature and precipitation trends: core_weather[["temperature_celsius", "temperature_fahrenheit"]].plot() plt.savefig("temperature_trends.png") # Export for reporting
   3. Train the Model Split data into training/testing sets and train the Ridge Regression model: reg = Ridge(alpha=0.1) reg.fit(train[prediction_parameters], train["target"])
   4. Evaluate Predictions Calculate MAE and visualize results: predictions = reg.predict(test[prediction_parameters]) print(f"MAE: {mean_absolute_error(test['target'], predictions)}") combined.plot().figure.savefig("predictions_vs_actuals.png")

4. Model Details Model Details Algorithm: Ridge Regression (L2 regularization). Features: 12 weather parameters (e.g., humidity, wind speed). Train-Test Split: Initial: First 35 rows for training, rows 36+ for testing. Validation: First 10 rows for training, rows 11+ for testing (in core_predictions function).

5. Results Mean Absolute Error (MAE): 8.585383009399957 Example: MAE ~8.5 degree celcius (lower is better). Visualizations: Temperature Trends Precipitation Trends Predictions vs. Actuals (predictions_vs_actuals.png).

6. Future Improvements Data: Use datetime index for proper time-series analysis. Add data imputation and normalization. Model: Test ARIMA, LSTM, or Gradient Boosting. Implement cross-validation. Evaluation: Track RMSE and R² metrics.

7. Contact For questions or feedback, contact: Name: Abhishek Sharma Email: abhisharma5sept2004@gmail.com Slack: PM Accelerator