Sweet orange is very much consumed by humans because oranges are rich in vitamin C, sweet oranges can be consumed directly to drink. The classification carried out to determine proper (good) and unfit (rotten) oranges still uses manual methods, This classification has several weaknesses, namely the existence of human visual limitations, is influenced by the psychological condition of the observations and takes a long time. One of the classification methods for sweet orange fruit with a computerized system the Convolutional Neural Network (CNN) is algorithm deep learning to the development of the Multilayer Perceptron (MLP) with 100 datasets of sweet orange images, the classification accuracy rate was 97.5184%. the classification was carried out, the result was 67.8221%. Testing of 10 citrus fruit images divided into 5 good citrus images and 5 rotten citrus images at 96% for training 92% for testing which were considered to have been able to classify the appropriateness of sweet orange fruit very well. The graph of the results of the accuracy testing is 0.92 or 92%. This result is quite good, for the RGB histogram display the orange image is good

Abstrack -The Covid-19 pandemic that has hit the world has changed the pattern of human life, including in the process of teaching and learning activities in universities. One of the universities affected by this pandemic is Pasir Pengaraian University. Lectures carried out by Pasir Pengaraian University during the Covid-19 pandemic consist of at least three forms, namely offline, online and blended learning. Efforts to assess which learning method is the most effective become important to measure the level of success of the teaching and learning process, so this study aims to determine the lecture strategy at Pasir Pengaraian University using the K-Means Clustering method. K-Means Clustering algorithm is a method in data mining that can be used for data grouping. CRISP-DM is a data mining methodology used in this study. The research dataset was obtained from the Even semester 2020 lecturer learning reports. RapidMiner was used as a tool to process the data. Clusters were formed as many as 3 (three) with the results of Cluster 1 (49 lecturers), Cluster 2 (17 lecturers), and Cluster 3 (54 lecturers). Based on these results, the lecture strategy with the Blended Learning type of learning is the most appropriate choice to be used at Pasir Pengaraian University, because apart from this Cluster having the highest number of memberships, in this Cluster the highest percentage of places to study are Classrooms/Labors and Meeting Applications, namely blend of offline and online lectures. The blended learning strategy has proven to be representative for use during the pandemic. Evaluation using DBI or Davies-Bouldin Index. The DBI value obtained is -1.163. Cluster evaluation is not good when viewed at this value, because it is negative and not close to zero.

Dalam upaya meningkatkan Usaha Kecil dan Menengah (UKM) di Kabupaten Rokan Hulu yang menjadi industri kreatif dan inovatif tentunya pendataan pesebaran UKM harus up to date dan valid sehingga pemerintah dapat memberikan kebijakan ataupun bantuan kepelaku usaha untuk mengembangkan usahanya apalagi dalam situasi pandemic ini. Penelitian ini bertujuan untuk mengelompokkan jenis UKM yang ada di Rokan Hulu menggunakan metode Fuzzy C-Means Clustering  dan membuat aplikasi baru berbasis Web untuk mendata persebaran UKM yang dilengkapi dengan peta pesebaran UKM . Fuzzy C-Mean Clustering (FCM) adalah suatu teknik pengclusteran data yang mana keberadaan tiap-tiap titik data dalam suatu cluster ditentukan oleh derajat keanggotaannya. Variabel yang digunakan berdasarkan omset, asset dan jumlah tenaga kerja. Sedangkan untuk pengelompokan jenis UKM dicluster menjadi 3 jenis, yaitu usaha menengah, usaha kecil dan usaha mikro. Berdasarkan hasil pengujian metode Fuzzy C-Mean Clustering dapat mengelompokkan jenis Usaha Kecil Menengah berdasarkan 3 cluster yaitu usaha menengah, usaha kecil dan usaha mikro, serta nilai validasinya rata-rata hampir mendekati angka 1, hal tersebut menunjukkan bahwa Fuzzy C-Means Clustering memiliki tingkat akurasi yang tinggi sebesar 80-90 %.

Tanaman pangan seperti padi masih menjadi perhatian dunia, termasuk di negara Indonesia khususnya Provinsi Riau. Kendala yang dihadapi Provinsi Riau adalah terbatasnya pengetahuan dalam pengelompokan produksi padi dan beras. Data mining memiliki beberapa algoritma dalam Clustering data, salah satunya K-Medoids. Produksi padi dan beras di Provinsi Riau dalam penelitian ini dikelompokkan menggunakan algoritma K-Medoids dengan menetapkan k  sejumlah 2 (dua). Perhitungan manual metode dan implementasi dalam RapidMiner untuk produksi padi dan beras dari tahun 2019 sampai 2021, maka diperoleh Cluster 1 (tinggi) sebanyak 4 kabupaten, sementara Cluster 2 (rendah) sebanyak 8 kabupaten. Hasil ini bisa menjadi informasi bagi pemerintah Provinsi Riau bahwa masih banyak kabupaten yang belum optimal dalam produksi padi dan beras, sehingga pemerintah bisa lebih fokus dalam meningkatkan produksi padi dan beras pada tahun-tahun berikutnya. Evaluasi Cluster menggunakan Davies-Bouldin Index (DBI) dan didapatkn nilai sebesar 0,626, sehingga bisa dikatakan bahwa evaluasi Cluster cukup baik, karena mendekati nol.

The purpose of pattern recognition is do the process of classifying an object into one particular class based on the pattern it has, so it can be used to recognize patterns of intestinal nematode worm types. One of the methods used in pattern recognition is by utilizing the artificial neural network method, the artificial neural network is able to represent a complex Input-Output relationship. For that the algorithm used is the perceptron algorithm. Perceptron is one method of Artificial Neural Networks. In the introduction of types of intestinal nematode worms, a computer must be trained in advance using training data and test data, this study discusses how a computer can recognize a pattern of types of intestinal nematode worms using the perceptron method. Based on the results of testing trials with input in the form of worm image scan results, based on the results of the perceptron method testing is able to recognize the pattern recognition of the types of intestinal nematode worms and be able to analyze with the right results of 100% for pinworms patterns, hookworm patterns, and 40- 50% for roundworms, by comparing the output value and the target value entered first.