Penerapan Metode CNN-LSTM Dalam Memprediksi Hujan Pada Wilayah Medan

Mhd. Alfandi (Universitas Budi Darma, Medan)
Pristiwanto Pristiwanto (Universitas Budi Darma, Medan)
A. M. Hatuaon Sihite (Universitas Budi Darma, Medan)

KOMIK (Konferensi Nasional Teknologi Informasi dan Komputer)
Vol 6, No 1 (2022): Challenge and Opportunity For Z Generation in Metaverse Era

The causative factor of rain can occur due to the air temperature in an area or also due to the volume of water carried by the clouds. Tomorrow's weather conditions are needed to draw up various plans. The people of medan city with the work of the majority of employees and traders need information about rainfall. For the past, rainfall forecasts depended heavily on the month, there was a dry season and a rainy season. But nowadays, rainfall is increasingly difficult to predict, so a model or system is needed that can accurately predict rainfall. In this study, it was explained about rainfall prediction using one of the ANN models to predict future rainfall called CNN-LSTM. CNN-LSTM is an artificial neural network system specifically designed to handle long-term time series data such as rainfall. In its architecture, the CNN-LSTM model uses 2 LSTM Hidden Layers consisting of 108 LSTM neurons in each layer. The activation function used is Tanh. The loss function used is Mean Square Error. The result obtained is a model that can better predict rainfall if the input data given to the model is getting longer which is marked by a smaller Root Mean Square Error value.