Design and Implementation of ML Based Temperature Forecasting Model for air conditioning using IoT

For accurate prediction of any physical parameter including ambient temperature by taking inputs from sensors, machine learning (ML) techniques are often preferred. In this paper, prediction and analysis of environmental real time temperature has been done using two ML models viz. Recurrent Neural Network (RNN) and Long Short Term Memory (LSTM). For analysis of collected data, traditional statistical methods such as Regression Analysis have been used. With the real time temperature sensor, data is collected periodically each day and then is used to train the ML models. During model development, various trails have been undertaken and the LSTM model has been found to be the most suitable. Various parameters of the model such as epochs, batch size have been further analyzed and with increase in the accuracy of the model, the predicted value is observed and a generalized performance is noted for real time use of the system. Further, the work involves the design of an internet of thing (IoT) set up as part of which the ML model works. Together the set-up becomes an intelligent IoT system which is used for remote monitoring through an android application. From experimental results the system has been found to be accurate.