This Assignment is worth 35% of the final module mark.

The challenge

Accurately   predicting   household   energy  consumption   allows   local   power   distribution companies to  better  forecast  energy  trends  and  perform  demand  management.  Power system demand management has gained heightened importance as the world transitions towards renewable energy. The rhetoric of the UK aiming to become “the Saudi Arabia of wind”  with the emergence of wind farms in the North Sea  has seen the nation pivot away from conventional fossil fuels towards cleaner, more sustainable sources. The North Sea's wind farms furnish a bountiful but highly variable power supply for UK households, providing a path towards national energy independence by reducing reliance on the importation of fossil fuels. Nevertheless, the primary technical hurdles hindering the increased adoption of wind energy in the UK revolve around efficiently transmitting power over long distances from the North Sea to urban centres, coupled with the challenge of seamlessly meeting demand during periods of low wind energy production or increased household energy use. In this project, we aim to address a component of these challenges by constructing a predictive  model for household energy demand. Our client, the national grid, may then use our model to help forecast when alternative energy production facilities need to be ramped up to meet household energy demands.

This  coursework  aims  to  create  an  effective   machine-learning  workflow  for   predicting household energy data. Your assigned tasks, detailed on the following page, require you to devise solutions independently. Alongside demonstrating your data modelling abilities, this assignment   evaluates   your   professional    engineering   skills,   including   adherence   to specifications, delivering tested and  commented  code,  meeting  client  requirements,  and justifying your approach.

Deliverables

1.  A report as a single PDF file;

2.   Code submitted as a single .zip file.

Data available

You  have  been  granted  access  to  the  'household_energy_data.csv'  dataset,  comprising 50,392  entries.  The  first  row  contains  the  names  of  each  feature  variable,  while  the subsequent  50,391   rows  contain  the  corresponding  data   points   associated  with  each household snapshot. These data snapshots capture household energy demands, smart meter readings of diverse household appliances, and concurrent weather conditions. The dataset consists of 30  columns,  each  representing  distinct features.  The  first  column  is  entitled “EnergyRequestedFromGrid_kW_ ” and this is the variable we are trying to predict.

Task/Assessment Description and Marks Available

Technical Report and Code.