Application of Multiple Linear Regression Method for Predicting Fish Production Based on Cultivation Type

Limbong, Hendra Putranta
Abdullah, Dahlan
Anshari, Said Fadlan

International Journal of Engineering, Science and Information Technology
Vol 5, No 1 (2025)

One of the contributors to Indonesia's economy is the fisheries sector, which has a high potential for development. Fisheries are a highly promising subsector for development in Indonesia's growth efforts. Based on data from the Central Bureau of Statistics of Dairi Regency, three types of aquacultures remain actively utilized in each subdistrict: ponds/freshwater ponds and paddy fields. This research aims to develop a fish production prediction system based on aquaculture types using the Multiple Linear Regression method. The accuracy of the prediction results will be measured using the Mean Absolute Percentage Error (MAPE). The results of this study indicate that in almost every subdistrict, especially pond aquaculture, the MAPE value is 20%, which means it has good accuracy. However, exceptions are found in the Siempat Nempu Hulu subdistrict, which has a MAPE value of 34.29%, and the Silahisabungan subdistrict, which has a MAPE value of 43.78%. Despite these values, they are still categorized as sufficient since they are 50%. The lower the MAPE value, the more accurate the prediction results. The findings of this research show that the multiple linear regression method can be considered correct. For future predictions, some results show negative values. For instance, in Silimapunggapungga subdistrict, a decline in production is predicted for 2024 with -114.779 tons and 2025 with -134.316 tons. The pessimistic prediction results are caused by the decrease in the X2 variable (area size), leading to a minor Y (production) value, potentially becoming negative if the contribution of X2 is no longer sufficient to balance the values of X1 (b1) and a. On the other hand, the Lae Parira subdistrict is predicted to experience an increase in production in 2024 by 87.024 tons and in 2025 by 84.380 tons. This system is implemented using the Python programming language. It is expected to help relevant stakeholders understand production trends and enhance the efficiency of fisheries resource management in the Dairi Regency.