fim

KO | EN

gitlite — search

#python #llm #rust #deep-learning #linux #docker #nodejs #macos #windows #awesome #cli #pytorch #mcp #ios

(★ 11)

fim is a collection of some popular frequent itemset mining algorithms implemented in Go.

fim 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# fim

[![Build Status](https://img.shields.io/github/workflow/status/paulfedorow/fim/Go)](https://github.com/paulfedorow/fim/actions/workflows/go.yml)
[![Go Report Card](https://goreportcard.com/badge/github.com/paulfedorow/fim)](https://goreportcard.com/report/paulfedorow/fim)
[![License](https://img.shields.io/github/license/paulfedorow/fim)](https://github.com/paulfedorow/fim/blob/main/LICENSE)

fim is a collection of some popular [frequent itemset mining](https://en.wikipedia.org/wiki/Frequent_pattern_discovery)
algorithms implemented in Go.

fim contains the implementations of the following algorithms:

* *apriori*: Based on [Fast Algorithms for Mining Associations Rules](https://dl.acm.org/doi/10.5555/645920.672836). As
  an optimization, the approach to determine the support of the candidates with a hash-tree and a transaction bitmap was
  replaced with an approach based on a [trie](https://en.wikipedia.org/wiki/Trie). As a further optimization, the
  generation of candidates of length 2 is omitted in favor of counting each occurring item pair in a 2D map.
* *eclat*: As described in [Scalable Algorithms for Association Mining](https://dl.acm.org/doi/10.1109/69.846291). The
  paper describes and analyses several approaches. The *eclat* algorithm is implemented based on the approach named
  **Eclat**. For the determination of frequent itemsets of length 2, a 2D map is used instead of a 2D array as described
  in the paper.
* *fpgrowth*: Based
  on [Mining Frequent Patterns without Candidate Generation](https://dl.acm.org/doi/10.1145/335191.335372). Implemented
  mostly as described in the paper. The only difference is that there is no special-casing implemented for FP-trees
  that are paths.
  
## Build

Execute the following commands to build the fim tool:

git clone https://github.com/paulfedorow/fim.git
    cd fim
    make

## Usage

To see which arguments fim supports execute the following command:

build/fim -h

The following example finds all frequent itemsets with a minimal support of 1% in the dataset contained
in `datasets/retail.dat`:

build/fim -a fpgrowth -i datasets/retail.dat -s 0.01

## Dataset format

fims dataset file format is as follows:

File = {Transaction}
    Transaction = Item {" " Item} "\n"
    Item = {"0" ... "9"}

Each line in the file is a transaction. A line is expected to be a series of integers separated by a single space. Each
integer is an item of the corresponding transaction.

## Performance

To determine which algorithm is the most efficient, the runtime of each algorithm was measured with different datasets
and decreasing minimal support. The datasets that were used are `retail.dat` and `chess.dat`
from [FIMI Dataset Repository](http://fimi.uantwerpen.be/data/). The datasets are respectively sparse and dense. The
results are shown below. The algorithm *fpgrowth* is best in terms of runtime.

![Runtime chart for datasets/retail.dat](https://raw.githubusercontent.com/paulfedorow/fim/main/charts/datasets_retail.dat.svg)

![Runtime chart for datasets/chess.dat](https://raw.githubusercontent.com/paulfedorow/fim/main/charts/datasets_chess.dat.svg)

fim

fim 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

fim

fim 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README