Predict the Percentage Error of LM35 Temperature Sensor Readings using Simple Linear Regression Analysis

Abstract

The development of embedded systems for monitoring is currently developing rapidly, often with the development of the Internet of Things (IoT). One of the monitoring systems that are widely designed is the temperature monitoring system. LM35 temperature sensors are often used in system design to read ambient temperature automatically. In the design results, the temperature data reading on the sensor has a percentage of data reading errors. In this study, we made an equation using simple linear regression to predict the percentage of error against the time that the sensor was used while reading temperature data. In the initial steps we designed a simple embedded system using the Arduino and the LM35 temperature sensor. Furthermore, measurements are made for a certain interval every five minutes. From the measurement data, the constant and regression coefficient are calculated by determining time as the dependent variable (Causal Factor Variable) and the percentage of error as the independent variable (Predictors). The results obtained a constant value of 5.56 and a regression coefficient of -0.014.