BROWSE
Detail
Conference Paper
A Study on the Data Analysis of Environmental Sensors Using Machine Learning Algorithms
Cited - time in 
Share 
Share
- Authors
- Kyong-Hee Lee, Hui Kim, Do-Hyun Kim
- Issue Date
- 2018-02
- Citation
- World Congress on Information Technology Applications and Services (World IT Congress) 2018, pp.1-4
- Language
- English
- Type
- Conference Paper
- Abstract
- Machine learning is currently being applied to prediction data, for example, for the classification of images, sentiment analysis, and linear regression. This paper suggests that for room occupancy detection, logistic regression and a recurrent neural network (RNN) be used to analyze variations in environmental sensor data. Classification and regression trees (CART), and logistic regression considered temperature, humidity, and CO2 for the estimation of occupancy. Notably, accuracy increased from 88% to 94% using logistic regression with a hidden layer. Furthermore, we introduced a recurrent neural network (RNN) to manage time series sensor data to predict trends of CO2 concentration with an RMSE value of 0.134.
- KSP Keywords
- Classification and regression tree(CART), Classification of Images, Co2 concentration, Data analysis, Environmental sensor, Hidden layer, Linear regression, Logistic Regression(LR), Machine Learning Algorithms, Occupancy detection, Recurrent Neural Network(RNN)