Machine Learning

Course project for Data-driven Building Energy Management

• Segmentation: from the ensemble of all the hourly powers acquired from all the appliances, find out the main patterns and infer the family lifestyle.
• Regression: using linear regression to further study the relationship between load pattern of each individual appliance and potential predictors such as temperature.

Competition on FeiFeng ShuChuang” 2018 Global IoT Data Competition

• Take comprehensive considerations of weather (precipitation, wind speed, wind direction, temperature) and civic emissions (NO2, SO2, Smoke) to be the factors influencing the air quality.
• Use data visualization to get intuitive information of all these factors and using Machine Learning algorithm (Random Forest) to obtain the importances of these features.

Click Here to see the details on GitHub

Course project for Urban System Modeling

• Electricity demand prediction within a specific region of the city with both Linear Regression and Bayes Ridge Regression models.
• Bayesian Network analysis for predicting the correlation of electricity consumption between regions within the city.