Published 2011 | Version v2
Dataset Open

Precipitation

Description

This dataset represents binary measurements of precipitation at 92 weather stations across four states in the United States of America. These four states are Illinois, Indiana, Iowa, and Missouri.

 

Task: The dataset can be used to study causal discovery algorithms as in Foygel Barber and Drton (2015).

 

Summary: 

  • Size of dataset: 371 x 93
  • Task: Causal Discovery Problem
  • Data Type: Binary Data
  • Dataset Scope: Standalone Dataset
  • Ground Truth: Known Graph
  • Temporal Structure: Time Series Data (irregular)
  • License: CC BY 4.0 (see https://www.ncei.noaa.gov/archive#v-pills-licensing)
  • Missing Values: No Missing Data

 

Missingness Statement: There are no missing values.

 

Features: The first column indicates the record time, wheras the following 92 columns represent the postal codes of the respective weather stations.

 

Files:

  • precipitation_dataset.csv: dataset
  • GroundTruth_delaunay.csv: Undirected graph obtained by applying Delaunay triangulation to the geographical layout via the postcodes.

Files

GroundTruth_delauney.csv

Files (77.3 kB)

Name Size Download all
md5:744fa6a36cfb3c16b7b7dfbfa8f56cb0
3.7 kB Preview Download
md5:11098cddc8e2c75bba57f0571abac488
73.6 kB Preview Download

Additional details

Related works

Is derived from
Data paper: 10.1214/15-EJS1012 (DOI)
Is published in
Journal article: 10.3334/cdiac/cli.ndp070 (DOI)

References

  • Menne, M. J., Williams Jr., C. N. and Vose, R. S. (2011). United States Historical Climatology Network Daily Temperature, Precipitation, and Snow Data
  • Rina Foygel Barber. Mathias Drton. "High-dimensional Ising model selection with Bayesian information criteria." Electron. J. Statist. 9 (1) 567 - 607, 2015. https://doi.org/10.1214/15-EJS1012