Unsupervised detection of botnet activities using frequent pattern tree mining

A botnet is a network of remotely-controlled infected computers that can send spam, spread viruses, or stage denial-of-service attacks, without the consent of the computer owners. Since the beginning of the 21st century, botnet activities have steadily increased, becoming one of the major concerns for Internet security. In fact, botnet activities are becoming more and more difficult to be detected, because they make use of Peer-to-Peer protocols (eMule, Torrent, Frostwire, Vuze, Skype and many others). To improve the detectability of botnet activities, this paper introduces the idea of association analysis in the field of data mining, and proposes a system to detect botnets based on the FP-growth (Frequent Pattern Tree) frequent item mining algorithm. The detection system is composed of three parts: packet collection processing, rule mining, and statistical analysis of rules. Its characteristic feature is the rule-based classification of different botnet behaviors in a fast and unsupervised fashion. The effectiveness of the approach is validated in a scenario with 11 Peer-to-Peer host PCs, 42063 Non-Peer-to-Peer host PCs, and 17 host PCs with three different botnet activities (Storm, Waledac and Zeus). The recognition accuracy of the proposed architecture is shown to be above 94%. The proposed method is shown to improve the results reported in literature.

Association Rule Integrasi Pendekatan Metode Custom Hashing dan Data Partitioning untuk Mempercepat Proses Pencarian Frekuensi Item-set pada Algoritma Apriori

Data mining dengan peran asosiasi sudah banyak digunakan oleh dunia usaha, salah satu algoritma yang sering digunakan untuk aturan asosiasi adalah apriori. Namun apriori memiliki kelemahan dalam hal performa, karena pada setiap penentuan frequent k-itemset harus melakukan scan database. Hal ini akan menjadi masalah apabila kandidat k-itemset memiliki dimensi yang banyak. proses scan database yang besar akan memakan waktu yang lama dan berpengaruh pada penggunaan memori dan prosesor. Apriori sudah sering dikembangkan, salah satu yang populer adalah Frequent Pattern (fp-growth), apriori dan fp-growth sama-sama merupakan algoritma untuk aturan asosiasi, hanya saja fp-growth menggunakan pendekatan yang berbeda dengan apriori yakni menggunakan pendekatan Frequent Pattern Tree (fp-tree). Meski fp-growth memiiki performa yang bagus ketika scan database namun rules yang di hasilkan oleh fp-growth tidak sebaik yang di hasilkan oleh apriori. Alternatif lain yang bisa digunakan adalah metode hashing, hal ini bisa menjadi solusi untuk mengatasi masalah dalam proses pencarian dan penentuan frequent k-itemset, sehingga proses scan database bisa lebih cepat. Tujuan penelitian adalah memperbaiki kinerja apriori dalam proses pencarian frekuensi itemset sehingga waktu scan database bisa lebih cepat

Data Mining Untuk Menganalisa Pola Pembelian Perak Dengan Menggunakan Algoritma Fp-Growth Pada Toko Emas Dan Perak Adi Saputra Tanjung

Pencatatan data transaksi pembelian perak harian pada took emas dan perak Adi Saputra Tanjung belum dilakukan dengan rapi dan data transaksinya dicatat ke dalam buku besar masih secara manual sehingga membuat pemilik toko kesulitan dalam menentukan barang apa saja yang laris di tokonya yang mengakibatkan promosi yang digunakan untuk meningkatkan penjualan di nilai kurang maksimal.Berdasarkan penelitian sebelumnya yang ditulis oleh Agus Nuryanto yaitu Penerapan Data Mining Menggunakan Algoritma Apriori Dan K-Means Untuk Meningkatkan Penjualan Toko Perhiasan Emas Setia Kawan, peneliti menganalisa pola pembelian perak untuk penemuan pola barang yang dibeli oleh pelanggan dengan harapan hasil penelitian dapat membantu rekomendasi promosi sehingga strategi pemasaran menjadi lebih tepat sasaran. algoritma yang digunakan adalah Frequent Pattern- Growth (FP-Growth) yaitu pengembangan dari metode Apriori yang merupakan salah satu alternatif untuk menentukan himpunan data yang paling sering muncul (frequent itemset) dalam sebuah kumpulan data dengan membangkitkan struktur data Tree atau disebut dengan Frequent Pattern Tree (FP-Tree).Hasil penelitian dari tahapan yang telah dilakukan, didapatkan nilai support sebesar 9% dan nilai confidence sebesar 30% dengan jenis perak yang dibeli konsumen yaitu cincin putar, mainan kalung, kalung nama, cincin rantai pilin dan anting. Hasilnya dapat membantu pemilik toko untuk mengambil keputusan dalam penentuan stok perak yang perlu diperbanyak sehingga meningkatkan keuntungan dan meminimalisir kerugian.

Implementation of K-Means Algorithm using Clustering Rules on Medical Data Sets

During the process of mining frequent item sets, when minimum support is little, the production of candidate sets is a kind of time-consuming and frequent operation in the mining algorithm. The K-Means algorithm does not need to produce the candidate sets, the database which provides the frequent item set is compressed to a frequent pattern tree (or FP tree), and frequent item set is mining by using of FP tree. These algorithms considered as efficient because of their compact structure and also for less generation of candidates itemsets compare to Apriori and Apriori like algorithms. Therefore this paper aims to presents a basic Concepts of some of the algorithms (K-Means Algorithmn, COFI-Tree, CT-PRO) based upon the FP- Tree like structure for mining the frequent item sets along with their capabilities and comparisons. Data mining implementation on spatial data to generate rules and patterns using Frequent Pattern (FP)-Growth algorithm is the major concern of this research study. We presented in this paper how data mining can apply on spatial data.

Extrication of Apriori Algorithm using Association Rules on Medical Data sets

During the process of mining frequent item sets, when minimum support is little, the production of candidate sets is a kind of time-consuming and frequent operation in the mining algorithm. The APRIORI growth algorithm does not need to produce the candidate sets, the database which provides the frequent item set is compressed to a frequent pattern tree (or APRIORI tree), and frequent item set is mining by using of APRIORI tree. These algorithms considered as efficient because of their compact structure and also for less generation of candidates item sets compare to Apriori and Apriori like algorithms. Therefore this paper aims to presents a basic Concepts of some of the algorithms (APRIORI-Growth, COFI-Tree, CT-PRO) based upon the APRIORI- Tree like structure for mining the frequent item sets along with their capabilities and comparisons. Data mining implementation on MEDICAL data to generate rules and patterns using Frequent Pattern (APRIORI)-Growth algorithm is the major concern of this research study. We presented in this paper how data mining can apply on MEDICAL data.

Determining Weighted, Utility-Based Time Variant Association Rules Using Frequent Pattern Tree

Introduction: The present research was conducted at Birla Institute of Technology, off Campus in Noida, India, in 2017.Methods: To assess the efficiency of the proposed approach for information mining a method and an algorithm were proposed for mining time-variant weighted, utility-based association rules using fp-tree.Results: A method is suggested to find association rules on time-oriented frequency-weighted, utility-based data, employing a hierarchy for pulling-out item-sets and establish their association.Conclusions: The dimensions adopted while developing the approach compressed a large time-variant dataset to a smaller data structure at the same time fp-tree was kept away from the repetitive dataset, which finally gave us a noteworthy advantage in articulations of time and memory use.Originality: In the current period, high utility recurrent-pattern pulling-out is one of the mainly noteworthy study areas in time-variant information mining due to its capability to account for the frequency rate of item-sets and assorted utility rates of every item-set. This research contributes to maintain it at a corresponding level, which ensures to avoid generating a big amount of candidate-sets, which ensures further development of less execution time and search spaces.Limitations: The research results demonstrated that the projected approach was efficient on tested datasets with pre-defined weight and utility calculations.