Prediction for a new year

[Marah35]

Hello,
I have an assignment to predict the unemployment rate in 2018, based on historical data.
The historical data attributes are the GDP number for 5 years and unemployment rate for 5 years.
The data as following:
Country GDP2013 GDP2014… unemployment 2013 unemployment…

I want to predict unemployment rate for 2018, and I am a new at KNIME.
anyone could help, please

[ScottF]

Sounds like a regression problem. Try the Linear Regression Learner node with unemployment rate as the target.

A side question for you: is using KNIME part of your assignment specifically, or just something you wanted to try on your own?

[mlauber71]

If it is not too complicated you could have a look at this Blog from KNIME. Prediction the future number of customers for a restaurant. You might be able to adapt this regression task and use the excellebt H2O.ai nodes in the process:

https://www.knime.com/blog/solving-a-kaggle-challenge-using-the-combined-power-of-knime-analytics-platform-h2o

[Marah35]

Its part of my assignment, and thank you so much for helping

[Marah35]

Thank you for helping.
Could you share with me a website that provide a free data that I can access.

[mlauber71]

The data of the restaurant challenge is hosted at Kaggle:

Recruit Restaurant Visitor Forecasting

(you might need to register an account)

And the KNIME workflow can be downloaded from the Examples Server

https://nodepit.com/download/workflow/public-server.knime.com%3A80%2F04_Analytics%2F15_H2O_Machine_Learning%2F07_Customer_prediction_with_H2O

https://nodepit.com/workflow/public-server.knime.com%3A80%2F04_Analytics%2F15_H2O_Machine_Learning%2F07_Customer_prediction_with_H2O

It could be that for your specific problem you would have to use some time series models. And you would have to think about you model setup. Do you have additional information about the countries or do you just have 2 numbers per county (GDP and unemployment). If so that would imply that you could create something like a general rule and that might be tricky but you could try. You might have to deal with outliers if a country had a very unusual year.

If you have the data you might share it so people could have a look.

[Marah35]

here is the data, I will first test it on small data.

And thank you for the link regarding restaurant’s prediction, I think I will test on it.
Thank you so much

[mlauber71]

Someone posted a similar question and it might be worth checking out this workflow:

How to predict a data in a future X year

I used an example from the KNIME Example server *1) to set up a workflow that demonstrates how this could be done. I am not yet satisfied with the graphics… The principle is that you use a time window of historical GDP data (Spain) for a time window of eg 5-10 years and then keep the last 2 (or 1 or 3) years as an unknown (to be predicted). I am not sure if you could use that to predict to the year 2048 that seems like a very far stretch. The models rely on historical GDP data from the years b…

You could try and adapt that not just with the GDP numbers but with the unemployment rate.

If this is for a educational assignment or something please try to read and understand the material that is given to you and try to understand if the examples provided do represent the theory. It might especially worthwhile to look into the effects the various models have on the prediction. There are more models out there and you might want to select some where you can then explain the effect they have.