chausner / arfftools Goto Github PK

In my code i need to know how many instances there are in a file (before enumerating them) to size some multi dimensional arrays.

As the methods get a reader as parameter, there is no way to re-open the reader.

So I added following two methods, to reset the ArffReader's stream and to return the instances count.

        /// <summary>
        /// Reset StreamReader to beginning.
        /// </summary>
        public void Reset()
        {
            streamReader.BaseStream.Position = 0;
            arffHeader = null;
        }

        /// <summary>
        /// Returns the Instance Count.
        /// </summary>
        /// <returns></returns>
        public Int64 Count()
        {
            Reset();
            ReadHeader();
            Int64 count = 0;
            while (ReadInstance() != null)
            {
                count += 1;
            }
            return count;
        }

Firstly, Thanks for creating this arff tool, it's awesome.

I am seeking a way to convert arff to csv format and I find out that on some large dataset it may costs a lot of time to load the entire dataset. For example, in the following dataset, it costs me about 1 hr when calling ReadAllInstances. FYI, my machine is a surface book 2 i7 32GB laptop.

• https://www.openml.org/d/41147

Is there any efficient way to make the loading speed much faster? Thanks