You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
DataFrameInternal - Using OrderedCollection over Array2D #105
DataFrameInternal currently uses Array2D (Previously it used Matrix #44)
Is there any specific reason such as speed/functionality for choosing Array2D?
Currently, while adding/removing a row, entire dataframe gets re-created. This becomes problematic for large data - eg: reading a csv file with thousands of rows results in calling addRow for every row in csv. DataFrameInternal is recreated for every such call.
I think using OrderedCollection would be better, since we can add elements at arbitrary indices. Are there any negatives for using OrderedCollection?
The way I was thinking of implementing this is having column-oriented OrderedCollections and contents will also be an OrderedCollection which holds these columns.
To access a row, fetch it's index, and iterate through contents, fetching index-th element for every column.
Will have to do a detailed performance profiling to see speed-downs in fetching row (if any), and speed-ups in adding rows.
Activity
AtharvaKhare commentedon Jun 21, 2019
The way I was thinking of implementing this is having column-oriented
OrderedCollections andcontentswill also be anOrderedCollectionwhich holds these columns.To access a row, fetch it's index, and iterate through
contents, fetching index-th element for every column.Will have to do a detailed performance profiling to see speed-downs in fetching row (if any), and speed-ups in adding rows.