Predicting energy consumption in 2025

[ghpook]

Please analyze my monthly energy consumption throughout years 2022 to 2024 and predict consumption for the year 2025.

I’m adding an Excel table, where

• Column A - month,
• Column B - year 2022 actual consumption
• Column C - year 2023 actual consumption
• Column D - year 2024 actual consumption

Please fill Column E with esitamted consumption data in year 2025.

tarbimine.xlsx (10.6 KB)

[rfeigel]

You have a very small data set to perform a credible regression or time series analysis. You can develop a model, but the output would be suspect.

[mlauber71]

@ghpook as @rfeigel has already said you might not be able to get a reliable production with so few data points so even an average might be a rough estimate.

Some energy consumptions might be quite stable but if it for example depends on using electricity for heating or cooling it will depend on the weather - a model would want such information. Also number of people in a household, time of vacations etc.

I tried my hand once at employment figures with the expected very mixed results. Maybe the example and discussions can help.

Can KNIME be used to show the employment and unemployment rate in the UK from the year 2008-2018, showing the prediction for the next year?

I set up an environment that first transforms you data from 2008-2016 into a table that has the Area (region) as an ID since I understand that is the unit you would want to predict. [image] The Target is taken from 2017 “Emp UK percent”. If you want Unemployment you would have to change that. The years are transformed into columns, while the first year of the data becomes _0 and the year before that _1 and so on. [image] Then I construct a Test dataset using the data of past 9 years from …

You can find more resources also about machine learning in general here

KNIME Snippets (4) — Start to Learn KNIME

[Iris]

I would disagree very slighty with the two above. While the data is too small for a real model, it is a very valid usecase to be able to do this prediction.

@ghpook I would recommend to first remove the seasonal effect, than use a simple model. This now depends on your knowledge about this. Either use a sliding window or a linear regression.
You would than need to reapply the seasonal effect.

[ghpook]

Thank you for the replies.
I have managed to solve my task with the help of another AI-tool, however, I will keep in mind the possibilities of this platform.

[rfeigel]

Against my better judgment, I developed a time series workflow employing fbprophet. Time Series Forecasting with fbProphet
It employs a Python environment which I’ve included in the workflow. It will propagate to your local machine if you have Conda installed. Read this:
Conda Environment Propagation – KNIME Community Hub
Here’s the workflow:
Energy Consumption Projection.knwf (162.7 KB)

Produces these graphs and tables. Note that the uncertainty is pretty large as would be expected from a long projection from sparse data.