top of page

Reduction of Air Con usage, an analysis using R & SQL

Software / programming languages used:
Excel
SQL


PowerBI
Statistical testing methods
Machine Learning

Project Details:

This project was built to learn the end-to-end process of data workflow from building the sensors that collect data to building the pipeline for the information to be stored and finally for the extraction, transformation and loading of the data to an analysis platform such as R and PowerBI.

​

From data that was collected and analysed, it was found that air conditioning usage could be reduced up to 40% in the given test period by ventilating a room well instead of turning the air con on. Suggestions were given on smart systems that could automatically turn the aircon off and open windows automatically using sensors and servos.

​

Scroll down to view project highlights and download links for project files.

​

This project has 2 other sub parts which can be accessed via these links:

​

Sensor uptime monitor using PowerBI Direct Query

​

Prediction model, A comparison of using Excel and R for machine learning / prediction

Highlights

Raspberry Pi, Arduino, ESP32

This was the hardware used. There were Arduino sensors, 3 WiFi sensors (pictured on the left) and 1 Ethernet based sensor (Pictured in the middle) which uploaded data every 10 seconds to a Raspberry Pi server running a MariaDB SQL server (Pictured on the right)

Temperature Predictive Analysis 1
Temperature Predictive Analysis 2
Temperature Predictive Analysis 3

Data was processed in SQL and analysed in R. Overall, 3 problem statements were proved and energy savings up to 40% could be achieved. An automated system that opens a room where air conditioning is used (which tends to be closed off from the outside) automatically when outside temperature is at an acceptable range whilst turning on the aircon during hotter temperatures could help to save electricity and greenhouse gas emissions. Please refer to the links below for the full report and project files.

bottom of page