Transformer-Based Household Electric Power Consumption Forecasting

Problem Statement

Develop a Transformer-based time-series forecasting model to predict household electric power consumption using historical electricity usage data. The model will use the previous 60 minutes of Global Active Power consumption to predict the consumption for the next minute. The model's forecasting performance will be evaluated using MAE and RMSE.

Dataset Description

Dataset: Individual Household Electric Power Consumption Dataset

The dataset contains measurements of electric power consumption from a household recorded at one-minute intervals.

Attribute	and Description

Date	- Date on which the measurement was recorded

Time-

Time of the measurement

Global_active_power-

Total active power consumed by the household in kilowatts (kW)

Global_reactive_power

- Household global reactive power

Voltage

- Average voltage measured in volts

Global_intensity

- Average current intensity in amperes

Sub_metering_1

- Energy consumption mainly associated with the kitchen

Sub_metering_2 -

Energy consumption mainly associated with the laundry room

Sub_metering_3

Energy consumption mainly associated with electric water heater and air conditioner

 

Variable Used in This Practical

Target variable: Global_active_power  

Download Dataset   

Go to Transformer-Based Household Electric Power Consumption Forecasting Model 

Previous 60 minutes of Global Active Power 

 ↓ 

 Transformer 

 ↓ 

 Global Active Power at next minute




For example:

Minutes 1–60 → Predict Minute 61