annoying to create a list and pass that to constructor. implement a set method for setting the hash function

I love python implementation of StreamSummary.
StreamSummary is usually done on incremental data.
e.g.
In my case, I am doing topK page views analysis via this StreamSummary code on google analytics of let's say past 1 month.
Now tomorrow I get a data of pageviews of tomorrow, I should be able to incrementally apply the script.

Please incorporate the save and load methods in the class.

save can save the necessary parameters and values in JSON.
load can reconstruct bucket and StreamSummary object from the loaded JSON.
This way, we can use it for incremental training.

Thanks

counter is never initialized!

stream summary is lacking a method to actually access the data, or otherwise make use of it. should probably provide a way to export the current elements in the stream, and test for existence in the dataset.

bloom array is based on BitArray, which recently had several critical bugfixes. these need to be ported into bloom_array

need to clean up allocated memory

running StreamSummary test program with test2.dat (uncommitted, but will commit later), size 3, yields the following assertion failure:

$ ./ss_test 3 test2.dat
ss_test: ../stream_summary.hpp:78: void Bucket::remove(const T&) [with T = std::basic_string]: Assertion `it != obj_map.end()' failed.
Aborted (core dumped)

need to convert text to markup

currently using tr1

see issue #7 for more info

The string representation __str as of now gives bucket number and members in that bucket.
Bucket number is not much of importance.

Please modify it to show
Bucket counter (instead of or in addition to) bucket number.

This way the user can know the counter of most frequently occurring items.

code is very inefficient. Should implement a binary tree for the bucket list so that inserts and searches are more efficient, or at least use binary search to locate elements (and then use a more efficient method to insert)

somehow the count min sketch commit failed to actually commit any code... need to recommit the file

Checking StreamingAlgorithms/KPS/kps.hpp...
[StreamingAlgorithms/KPS/kps.hpp:76] -> [StreamingAlgorithms/KPS/kps.hpp:79]: (error) Iterator 'map_it' used after element has been erased.

in stream summary (and perhaps others), insertions into the STL containers can fail. the returned pair needs to be checked. if false, error needs to be thrown