Table 2.

Characteristics of the Dataset Repository Corpus Used in This Study

Type	Characteristics	Mean (±SD)	Quantile
Data file	no. of rows (csv)	4,115 $\left(\pm \mathrm{50,094}\right)$	[39.0, 92.0, 108.0]
	no. of columns (csv)	20.5 $\left(\pm 373\right)$	[3.0, 5.0, 12.0]
	no. of rows (xls(x))	607 $\left(\pm 13610\right)$	[28.0, 65.0, 108.0]
	no. of columns (xls(x))	30.5 $\left(\pm 412.1\right)$	[8.0, 15.0, 19.0]
	no. of missing values csv (ratio)	8.9 $\left(\pm 17.5\right)$	[0.0, 0.0, 11.5]
	avgerage size of data files (csv)	331,343 $\left(\pm \mathrm{3,719,328}\right)$	[1,625.0, 8,375.0, 47,752.5]
	average size of data files (xlsx)	428,586$\left(\pm \mathrm{2,595,222}\right)$	[18,804.0, 34,723.0, 121,633.0]
Repository	size of repository	51,372 kilobytes $\left(\pm \mathrm{211,729}\right)$	[983.0, 7,740.0, 32,715.0]
	no. of open issues	5.2 $\left(\pm 51.2\right)$	[0.0, 0.0, 0.0]
	no. of closed issues	40.6 $\left(\pm 552.3\right)$	[0.0, 0.0, 2.0]
	description length	7.2 $\left(\pm 9.2\right)$	[1.0, 5.0, 10.0]
	ratio of data files per repo	7.2% $\left(\pm 13\text{\%}\right)$	[0.3, 1.9, 8.0]
	age of repository (days)	1,521.9 $\left(\pm 539.7\right)$	[1,108.0, 1,478.0, 1,844.0]
	ratio of problematic files with respect to a standard config (Pandas)	0.3% $\left(\pm 2.6\text{\%}\right)$	[0.0, 0.0, 0.0]
README	no. of words in README (non-code related)	378.2% $\left(\pm \mathrm{1,126.6}\text{\%}\right)$	[10.0, 112.0, 431.0]
	no. of tables	0.1 $\left(\pm 1.0\right)$	[0.0, 0.0, 0.0]
	no. of code blocks	1.4 $\left(\pm 4.7\right)$	[0.0, 0.0, 1.0]
	no. of headers	3.6 $\left(\pm 17.0\right)$	[1.0, 1.0, 5.0]
	no. of urls	9.1 $\left(\pm 36.9\right)$	[1.0, 3.0, 12.0]
	no. of images	0.7 $\left(\pm 3.9\right)$	[0.0, 0.0, 0.0]

Average values are reported in the “mean (±SD)” format. Quantiles values are reported in the [$x_{25}$, $x_{50}$, $x_{75}$] format, where $x_{25}$, $x_{50}$ and $x_{75}$ represent the 25th, 50th, and 75th quantile of a particular group's characteristic.