Weather Data Analysis with Partial Least Squares Regression

This code provides a systematic approach to handling and analyzing large-scale weather datasets stored in the .nc format.

See the article in the description for more informations.

Table of Contents

• Prerequisites
• Execution Overview
• Steps
  • Configuration

Prerequisites

Ensure you have the following Python packages installed:

os netCDF4 numpy pandas datetime sklearn xarray multiprocessing tqdm joblib

Execution Overview

The script performs the following operations:

1. Loads data from .nc files located in a specified directory.
2. Processes and merges these datasets.
3. Validates the time range of data.
4. Splits the data into training and testing datasets.
5. Applies Partial Least Squares (PLS) Regression.
6. Transforms and validates the resultant data.
7. Finds the most similar data points using parallel computation.

Steps

Configuration

time_interval_reconstruction = ["2018-01-01 00:00:00", "2019-12-31 23:54:00"]
predict_variable = 'WSPDchyv2_2011_2019'
k = 3
folder_path = '/Users/Murilo/weather_data'
netCDF_resolution = 1