K-MEANS ALGORITHM TO DETERMINE MARKETING STRATEGY AT CODEVERSE COMPUTER ACCESSORIES STORE

: Artificial Intelligence (AI) is currently gaining popularity across various industries, including healthcare, finance, and others. In this study, AI technology is employed to devise an optimal marketing strategy for Code Verse Computer Accessories Store using the K-Means algorithm. As part of machine learning, the K-Means algorithm, categorized under unsupervised learning, is implemented to cluster sales data for computer accessory products over the last three months of 2023. The results of the K-Means analysis identify two main clusters. Cluster one (Cluster 1) comprises products such as Mouse, Keyboard, Monitor, Headset, and Speaker, indicating consistent purchasing patterns and high consumer interest. Recommendations are made to increase stock for Cluster 1. Meanwhile, Cluster two (Cluster 2) consists of Mic products with lower interest, and it is not advisable to increase stock. The implementation of K-Means provides insights into purchasing patterns, enabling Code Verse to develop more effective marketing and inventory management strategies.


INTRODUCTION
Artificial Intelligence (AI) is currently a very popular technology, and it has been adopted by various industries, ranging from the healthcare sector to finance, and so on [1].The use of AI in the healthcare industry encompasses a wide range of applications, such as diagnostic assistance through medical image analysis, prediction of potential health problems through patient data 396 analysis algorithms, accelerated drug discovery, personalized treatment through genetic data analysis, administrative support with AI systems, and AI-based virtual health assistant services to provide 24/7 access to patients [2].All of these optimize patient care, improve operational efficiency, and provide innovative solutions in the healthcare industry [2].
The  Implementing the K-Means algori thm strategically enhances CodeVerse software store's marketing efforts, consolidating its leading position as a one-stop destination for computer users seeking integrated solutions, spanning high-quality hardware sales to professio nal computer repair services.

METHOD
In this research method, resear chers use several stages.These stages will be described in Image 1.

Data Collection
The dataset consists of 1000 entries of sales data from October to December 2023 at CodeVerse Computer.
Then for the explanation of the above formula is as follows: d (x-y) : the distance between the data at points x and y x: object data point y: centroid data point I: number of data attributes The K-Means method, widely applicable in tasks like image and text analysis, exhibits drawbacks due to its initial random centroid distribution, often resulting in varied outcomes and nonuniform classification, leading to inconsistencies in results.

Data Mining
The basic principle of data mining is to discover hidden information in databases, which is an important part of Knowledge Discovery in Databases (KDD), which aims to find useful information and patterns in data [13].

Image 2. Data mining working order
In Image 2, the data mining process begins with understanding the relevant domain or application, followed by careful selection, cleaning, and conversion of datasets.Subsequently, the data is reduced and subjected to prediction processes to extract pertinent components, which are then formulated into regression variables for specific analysis.[14].

K-Means Algorithm
K-Means is an unsupervised method that selects a subset of the population as the centers of the first group [15].

Clustering
Basically, clustering is an attempt to find and group data that has something in common [16].Clustering technique is an approach in the field of data mining that groups a number of data into clusters based on similar characteristics or objects [17].

Cluster Number (K) Selection
At this point, the researchers identify the cluster centers, in this case represented by statements 1 and 5. Table 1.data from October to December.Based on table 1 data, the number of clusters was determined.The cluster data used can then be seen in table 2.

Centroid Initialization
Subsequently, centroid calcula tion will be performed using the formula, with Table 3 providing the initial centroids K1 and K2, where K1 denotes the selling products and K2 represents products that are not selling well.After K1 and K2 are determined randomly, the next step is to carry out calculations to get the values of C1 and C2.In this calculation, the Euclidean distance formula is used.Further information can be found in Table 4.

Data Grouping
The results of data grouping can be seen in table 5. From the data in table 5, it can be concluded that data 1, 2, 3, 4, and 5 are included in Cluster 1.Meanwhile, data number 6 is included in Cluster 2.

Centroid Update
In this research, centroids were recalculated based on existing data points in cluster table 5.Then, for conclusions the centroid update results can be seen in table 6.To find the values in table 6, the calculations can be seen in manual calculations as follows.
( 2 ) ( 3 ) ( 4 ) ( 5 ) Reflection is carried out until each cluster is the same as the previous iteration, in this case, those in Iteration 1.For the new data grouping, it can be seen in table 7.
1st to May 31st, 2023, using the K-Means Clustering Algorithm on 949 records.The analysis revealed three clusters: low, moderate, and high.Cluster 1, predominantly comprising males aged 40 and above, exhibited a significant incidence of Acute Respiratory Tract Infections (ARTIs), indicating potential health disparities in the community.[11].The essence of this study is to use the K-Means clustering algorithm to analyze the sample of Riau University scholarship recipients in 2020, 2021 and 2022.The research results show that group 0 is the majority of students from the D3 Commercial Shipping Manageme nt research program at level 5. Currently, most Cluster 1 students from the seventh semester of the accounting and management studies program, with a GPA greater than or equal to 3.51, will graduate.In addition, group 2 is dominated by students of the nursing education program of the fifth week with the lowest GPA of 3.51 [12].
the flow of the research, starting from data collection, data transformation, the use of the K-Means algorithm, then the results of the research appear.

Table 2 .
Sales For Data Cluster Materials

Table 4 .
Cluster data calculations

Table 5 .
grouping data results

Table 6 .
data cluster after update