Julia: Jurnal Ilmu Komputer An Nuur

ANALISIS SENTIMEN PADA TWITTER TENTANG ISU PERILAKU ANTISOSIAL DENGAN ALGORITMA NAÏVE BAYES

2024-01-06T10:39:02+07:00

In 2023, around 78.19% of the 275.77% or 215.63 million Indonesian population will be connected to the internet, with positive impacts such as fast communication, entertainment and new knowledge. The internet makes non-cash transactions easier and has negative impacts such as addiction and antisocial behavior such as indifference to people around you. Teenagers often access social media, especially Twitter, to express opinions and vent both positive and negative. Sentiment analysis is used to determine opinions about antisocial behavior on Twitter by using text mining techniques to analyze teenagers' opinions. Naive Bayes and SVM algorithms are used in sentiment analysis on the Twitter dataset to analyze antisocial behavior. Actions to evaluate the Naive Bayes algorithm in assessing antisocial behavior sentiments had the best accuracy results of 59.71% with k=7 without n-grams. The Naïve Bayes algorithm with k=5 and n-gram n=2 has the best precision of 33.76% and the best recall of 33.45%. Future research can try to use other classification algorithms such as KNN, SVM, etc. To find the best accuracy of the antisocial behavior dataset.

Keywords: Internet; Twitter; antisocial behaviour; Sentiment analysis;

IMPLEMENTASI ALGORITMA FP-GROWTH UNTUK REKOMENDASI PRODUK DI TOKO LM MART

2024-01-04T19:10:35+07:00

LM Mart is one of the BumDesa shop businesses located on Jl Raya Purwodadi-Semarang Km.13, Godong sub-district, Grobogan Regency. The products sold include various basic food items (nine basic commodities) for general community needs. Data is stored in the LM Mart store database. One of them is increasing transaction data. With the increasing volume of data at LM Mart, the analyst's function of analyzing data manually must be replaced by computer-based applications. The problem with the LM Mart Store is that traders lack the ability to observe consumers' desires and needs, which of course will have an impact on increasing product sales. Besides that, sales transaction data, if processed, can produce useful information which can become a sales strategy to improve marketing. The FP-Growth algorithm will be used for the association approach in this research. The FP-Growth algorithm is a development of the apriori algorithm, it corrects the shortcomings of the apriori algorithm. To obtain a frequent item set, the a priori algorithm must generate candidates. From the research results, calculations using RapidMiner with a Support value of 30% and a Confidance value of 80% with transaction data of 800 records produced 36 rules.