Datasets:

Modalities:
Tabular
Formats:
parquet
ArXiv:
Libraries:
Datasets
pandas
License:
chaenykim commited on
Commit
538474c
1 Parent(s): 2402bef

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +24 -0
README.md CHANGED
@@ -61,3 +61,27 @@ configs:
61
  - split: test
62
  path: data/test-*
63
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
61
  - split: test
62
  path: data/test-*
63
  ---
64
+
65
+ ## Dataset Structure
66
+
67
+ ### Data Fields
68
+
69
+ - **object_id**: unique object identifier
70
+ - **times_wv**: 2D array of shape (N, 2) containing the observation times (modified Julian days, MJD) and filter (wavelength) for each observation, N=number of observations
71
+ - **target**: 2D array of shape (N, 2) containing the flux (arbitrary units) and flux error for each observation
72
+ - **label**: integer representing the class of the object
73
+ - **redshift**: true redshift of the object
74
+
75
+ ### Data Splits
76
+
77
+ - **train**: 90% of the PLAsTiCC training set
78
+ - **validation**: 10% of the PLAsTiCC training set
79
+ - **test**: full PLAsTiCC test set
80
+
81
+ ## Usage
82
+
83
+ ```python
84
+ from datasets import load_dataset
85
+ # Loading the data
86
+ dataset = load_dataset("BrachioLab/supernova-timeseries")
87
+ ```