Energy Consumption Prediction Analysis

Sandeep Marshall presents a project on predicting household energy consumption using sensor and weather data. The project aims to balance electricity supply and demand to prevent blackouts. Data from a Belgian building is analyzed, focusing on temperature and humidity. Feature engineering and PCA reduce data dimensions. Various models are trained, with Extra Trees Regressor performing best. The analysis highlights the importance of time zones in energy consumption. Conclusions emphasize the effectiveness of tree-based models and PCA in handling data with low correlation to target variables.

Why is it important to predict energy consumption in households?

• To reduce electricity bills

• To improve the quality of air in homes

• To prevent blackouts by balancing supply and demand

• To increase the lifespan of electrical appliances

What type of data was collected from the building in Belgium?

• Financial data

• Sensor data including temperature and humidity

• Traffic data

• Social media data

What is the target variable in the dataset?

• Temperature levels

• Humidity levels

• Total energy consumption

• Wind speed

What is the purpose of using PCA in this study?

• To improve data collection methods

• To increase the number of features

• To reduce dimensionality while maintaining variance

• To eliminate all correlations

Which time of day showed significant correlation with energy consumption?

• Midnight

• Noon

• Afternoon

• Morning and evening

Which model performed best in predicting energy consumption?

• Linear Regression

• Support Vector Regression

• Extra Trees Regressor

• K-Nearest Neighbors

How did PCA affect the model's performance?

• It improved the performance drastically

• It made the model unstable

• It had no significant effect

• It significantly worsened the performance

What was the R2 score of the best-performing model?

• 0.800

• 0.900

• 0.645

• 0.500

Which feature was found to have significant importance in the model?

• Temperature inside

• Session time

• Wind speed

• Humidity outside

What conclusion was drawn about the relationship between temperature, humidity, and energy consumption?

• They have a strong positive correlation

• They have little to no correlation with energy consumption

• They are irrelevant to the study

• They are the most important predictors