seiflotfy

10 ms·

61.
by seiflotfy 9y ago
2 different things. This reduces the size of the HyperLogLog the other improves the accuracy.
62.
by seiflotfy 9y ago
In my test metro hash gave me higher throughput, must be something with the xxhash implementation in Go
63.
by seiflotfy 9y ago
Author here, the main 3 things that make this stand out is: 1) 4-bit register instead of 5 (HLL) and 6 (HLL++), but most implementations use 1 byte registers out of convenience 2) Sparse representation for lower cadinalities (like HyperLogL
64.
Show HN: An improved version of HyperLogLog(github.com)
194 points by seiflotfy 9y ago|44 comments
65.
Show HN: Stream cardinality estimation since any timestamp (mod. HyperLogLog)(github.com)
1 points by seiflotfy 9y ago|0 comments
66.
Show HN: Time Adaptive Count-Min Sketches for Summarizing Data Streams(github.com)
1 points by seiflotfy 9y ago|0 comments
67.
Show HN: RLogLog-Beta - Cardinality estimation since a given timestamp(github.com)
3 points by seiflotfy 9y ago|0 comments
68.
Show HN: A better bias estimation for HyperLogLog(github.com)
3 points by seiflotfy 10y ago|0 comments
69.
Show HN: LogLog-Beta: Like (slightly better than) HyperLogLog++ but simpler(github.com)
7 points by seiflotfy 10y ago|0 comments
70.
Show HN: HyperBitBit - aiming to beat HyperLoglog (initial implementation)(github.com)
4 points by seiflotfy 10y ago|0 comments
71.
Show HN: Count-Min-Log 4 Golang == better than Count-Min Sketch(github.com)
3 points by seiflotfy 10y ago|0 comments
72.
Show HN: Initial Microsoft Bing Speech API for Go (In Progress)(github.com)
10 points by seiflotfy 10y ago|2 comments
73.
Show HN: Playing with net dotnet and IronFunctions(seif.codes)
7 points by seiflotfy 10y ago|0 comments
74.
Rustifying IronFunctions(seif.codes)
4 points by seiflotfy 10y ago|0 comments
75.
Importing AWS Lambda to IronFunctions(seif.codes)
2 points by seiflotfy 10y ago|0 comments
76.
by seiflotfy 10y ago
you can import your AWS Java code using the command line https://github.com/iron-io/functions/blob/master/docs/lambda...
77.
by seiflotfy 11y ago
Sure thing. So imagine you have a domain with a expected cardinality (which you pass) of 1 million and a threshold of 10k. What will happen is every time you insert an event it will be multiplexed into all 4 sketches. Each one of them will
78.
by seiflotfy 11y ago
Redis provides a bitset that can be used as a bloom filter. You'd end up having to hash k times on a client side, while also calculating the ideal size of the bitset for the expected "n" values and "e" error rate. T
79.
Show HN: Skizze – High-throughput probabilistic data structure service(github.com)
53 points by seiflotfy 11y ago|6 comments
80.
Skizze: Second Alpha Release (Probabilistic Data-Structures Storage)(njp.io)
10 points by seiflotfy 11y ago|0 comments
81.
Skizze – A probabilistic data-structures service and storage (Alpha)(geekyogre.com)
4 points by seiflotfy 11y ago|0 comments
82.
S-Bitmap: Distinct Counting with a Self-Learning Bitmap (like HyperLogLog)(github.com)
1 points by seiflotfy 11y ago|0 comments
83.
by seiflotfy 11y ago
With zipf distribution did you mean for every flow insert a zipf fetched number of events?
84.
by seiflotfy 11y ago
voila. my first benchmarks based on my use case https://news.ycombinator.com/item?id=10149522
85.
Semi Evaluation Count-Min Sketch, Count-Min-Log and PMC (based on Go Code)(geekyogre.com)
2 points by seiflotfy 11y ago|0 comments
86.
by seiflotfy 11y ago
I am working on a benchmarking blog post where I will be comparing count-min, count-mean-min, count-min-log and psc... Its very nice to see how much space is being saved sadly estimating is a bit expensive (yet doable). I plan to also do so
87.
Probabilistic Multiplicity Counting(github.com)
16 points by seiflotfy 11y ago|7 comments
88.
Count-Min-Log sketch: Approximately counting with approximate counters(github.com)
4 points by seiflotfy 11y ago|0 comments
89.
Rust-Farmhash: Implementation of farmhash in rust(github.com)
2 points by seiflotfy 11y ago|0 comments