Improving Temperature Sensor Accuracy in the IoT Trainer Kit by Linear Regression Method

The rapid development of Internet of Things (IoT) makes people in higher education must train their students to be better prepared in advancing and implementing that topic. Therefore, to improve student comprehension, we need tools such as trainer kit as a learning media. The IoT Trainer Kit has been created in Bandung State Polytechnic called the I-Kit which has many features. Inputs include DHT 11 temperature, humidity sensors and RFID. The controller used is Arduino Nano. Output for features in the trainer kit will appear on the web page. This I-Kit also has several communication devices such as Bluetooth, LoRa, ESP 8266 and SIM 800. However, before using the I-Kit as a learning medium, we must make the features in this trainer kit precision first. But the training kit is also not necessarily reliable, it must be tested and improved for the performance of its features. In this paper, we have improved accuracy of the DHT 11 temperature sensor on the I-Kit. Improvement was carried out using the linear regression method, to find out the correlation between The temperature of the thermometer with the temperature read on the sensor in the trainer's kit. Then this regression equation is entered into the temperature program in Arduino. When comparison is made between error and deviation standard before and after doing regression, the error rate is decrease byd 80.9 % from 7.3 become 1.39. The deviation standard which represents tolerance from sensor decrease 20% from 0.88 becomes 0.704.