What does it do?

• 

  : )

Details

Objective : to produce the best understanding of customer load using machine learning

1. Produce the most accurate forecast of the total load and the load per cluster for the next 7 days.
2. From the load data, research “load disaggregation” and break down the loads for each customer into different parts. Most imaginative breakup of the load wins - but it must be done using code Details: • Load data is provided at 15 minute intervals. You have the consumption in W. This is the total electricity that each house is using, adding up all of the appliances. • There is some missing and erroneous data, a good sort of the data is needed. • Forecasts: • You have some of the data we think is needed to produce good forecasts, but not all. You can also do some reading on what other data can make a good forecast and go out and find it yourself - or to talk to us about it. • There are loads of Machine Learning methods available for load forecasts (e.g. search NILM). Think about what type of model you might need before committing time programming it up. • There’s lots of information online on load desegregation. • Tip 1: you need to think how detailed you want to go - down to individual appliances or sticking to base load and peak load. There is a balance between what is achievable and what is ambitious enough. • Tip 2: New Zealand houses generally have a hot water tank which is heated up with lots of power once or twice a day. They also have a heat pump for heating in the winter (May - Sept) and possibly for cooling in the summer (Nov-Feb)## Inspiration

What it does

It predicts the data for the next 7 days.

How we built it

By not sleeping for two days straight and learning a lot about data sciencey stuff.

Challenges we ran into

Hunger, Lack of Sleep

Accomplishments that we're proud of

Graphs and Forecasts! Great Success!

What we learned

Data Sciencey Stuff!

What's next for What does it do?

Adapt, Improve, Overcome

Built With

• jupyternotebook
• python