public double GetMultiColumnEntropy(DataTable table, params Column[] columns)
{
try
{
IEscaper escaper = table.Database.Escaper;
//Select statement is actually just "SELECT SUM(pcount * log10({2} / pcount)) / (log10(2) * {2}) "
//Obfuscated due to postgresql different log function names.
//The .0 after the first {2} is to force float type
string sourceQuery = @"SELECT " + escaper.RoundFunction("SUM(pcount * " + escaper.Log10Function("{2}.0 / pcount") + @") / ( " + escaper.Log10Function("2") + @" * {2})", 10) +
@"FROM (SELECT (Count(*)) as pcount
FROM {0}
GROUP BY {1}) exp1";
StringBuilder columnsString = new StringBuilder(columns.Length * 10);
bool first = true;
foreach (var column in columns)
{
var escapedColumn = escaper.Escape(column);
if (first)
{
columnsString.Append(escapedColumn);
first = false;
}
else
{
columnsString.AppendFormat(", {0}", escapedColumn);
}
}
var query = string.Format(sourceQuery, escaper.Escape(table), columnsString.ToString(), table.Cardinality);
var res = table.QueryTable(query);
if (res is DBNull)
{
return(0);
}
else
{
return(Math.Max(0, Convert.ToDouble(res)));
}
}
catch (Exception ex)
{
Console.Error.WriteLine(ex.Message);
return(0);
}
}
/// <summary>
/// Escape
/// </summary>
public static string Escape(this string value, IEscaper escaper)
{
if (escaper is null)
{
throw new ArgumentNullException(nameof(escaper));
}
return(escaper.Escape(value));
}