Kafka based LSTM model for streaming data prediction
S. Geetha, V. Kalaivani · AIP conference proceedings · 2022
In the current technological world, the data is getting generated abundantly through various sources. The time dependent data generated through sensors is one among them. The most of the time dependent or time series data is streaming in real-time onfixed flow rates and sometimes, it will be on variation. When there is a change or no change in the flow of data, the existing state of the art prediction models provides low accuracy of prediction. Hence, it is proposed to have better accuracy of prediction on the streaming data with LSTM and Kafka Framework. The Kafka-LSTM model performed much better than the other conventional models. The models were evaluated using mean absolute error and root mean square error. The proposed method performs with better accuracy and reduced error rate than the other conventional models.