I would like to turn:
DateTime ColumnName Min Avg Max
2012-10-14 11:29:23.810000 Percent_Used 24 24 24
2012-10-14 11:29:23.810000 Current_Count 254503 254503 254503
2012-10-14 11:29:23.810000 Max 1048576 1048576 1048576
2012-10-14 11:34:23.813000 Percent_Used 24 24 24
2012-10-14 11:34:23.813000 Current_Count 254116 254116 254116
2012-10-14 11:34:23.813000 Max 1048576 1048576 1048576
Into a dataframe where the the DateTimes are unique (an index) and the columns are:
DataTime, Percent_Used_Min, Percent_Used_Avg, Percent_Used_Max, Current_Count_Min, Current_Count_Avg, Current_Count_Max, Max_Min, Max_Avg, Max_Max
Basically, I want to mimic R's melt/cast without getting into hierarchical indexing or stacked dataframes. I can't seem to to get exactly the above playing with stack/unstack, melt, or pivot/pivot_table -- Is there a good way to do this?
As An example, in R it would be something like:
dynamic_melt = melt(dynamic, id = c("DateTime", "ColumnName"))
recast = data.frame(cast(dynamic_melt, DateTime ~ ...))
The above data will be variable (i.e. the values of ColumnName won't always be the same thing, there might be more or less of them, and different names).