Skip to content

Latest commit

 

History

History
70 lines (54 loc) · 2.51 KB

README.md

File metadata and controls

70 lines (54 loc) · 2.51 KB
Raw

Global Temperature Prediction and Analysis

Overview

This project aims to investigate temperature changes over time and predict future temperature patterns on a regional and global scale. We employ time series forecasting methods, including neural networks, ARIMA, and SARIMAX, using the GISTEMP v4 dataset from NASA.

Table of Contents

Project Structure

  • global_temp_nn/: Jupyter notebooks for data exploration, cleaning, analysis, and neural network implementation.
  • arima_sarima/: Jupyter notebooks for ARIMA and SARIMAX time series forecasting.
  • images/: Contains images generated during the analysis.
  • data/: Dataset files used in the project.
  • README.md: Project overview and instructions.
  • temperature_video_june.avi: Temperature anomalies plotted on a world map from 1880 to 2023.

Video

temperature_video_june

Data

The dataset used in this project includes:

  • Global Surface Temperatures
  • Northern Hemisphere Temperatures
  • Southern Hemisphere Temperatures
  • Zonal Temperatures
  • Extended Reconstruction SSTs Version 5 (ERSSTv5) (NetCDF file)

Analysis and Modeling

  • Exploratory Data Analysis (EDA): Initial exploration of the dataset to understand patterns and trends.
  • Data Cleaning: Handling missing values, interpolation, and ensuring data integrity.
  • Time Series Forecasting: Utilizing ARIMA and SARIMAX for time series forecasting.
  • Neural Network Models: Implementing neural networks for more complex analyses.

Results

  • Visualizations of temperature anomalies over time.
  • Forecasts of future temperature trends using ARIMA and SARIMAX.
  • Neural network predictions for complex analysis of temperature patterns.

How to Run

Download the zip file and run the individual Jupyter notebooks:

  1. global_temp_nn.ipynb
  2. arima_sarimax.ipynb

Dependencies

  • TensorFlow
  • Statsmodels

Install the required packages using:

pip install tensorflow statsmodels

Contributing

Shubham Garg- sg8311@nyu.edu

Raunak Shukla - rs8668@nyu.edu

Phani Varma Gadiraju - pg2542@nyu.edu

License

This project is licensed under the MIT License.