Real world demand dataset

Question

I am looking for a publicly available real world demand data to compare the performance of some algorithms. The dataset should include some products with data of a sufficiently long period of time. I know the FoodMart dataset, whose structure you can find described on this page; but, this dataset includes a small number of records per product so that it is not suitable for my purpose.

I appreciate any link to such dataset or papers which have made their dataset publicly available.

give a try to the google dataset search engine in this address: https://toolbox.google.com/datasetsearch — Oguz Toragay, Aug 16 '19 at 15:27
Kaggle has lots of datasets containing demand data, e.g., here. They are usually designed for forecasting/prediction, but I don't see why you can't use them for other purposes. — Ryan Cory-Wright, Aug 16 '19 at 17:50
The time series data library from Rob Hyndman has hundreds (~648) of time-series data. It depends on what you call "demand," but I'm not sure how many meet your "sufficiently long period of time" criterion. — SecretAgentMan, Aug 16 '19 at 18:37
@OguzToragay Thanks !. I know the search engines. My question was mostly about a given known database that you are aware of. — Afshin Oroojlooy, Aug 17 '19 at 05:58
@RyanCory-Wright Thanks!. I am looking into the mentioned dataset. The problem with Kaggle is that there are a lot of dataset, and it might be time-consuming to understand and analyze each dataset to find a proper one. — Afshin Oroojlooy, Aug 17 '19 at 06:02
@SecretAgentMan, Thanks. This looks fantastic. But, I could not find any description of each dataset. DO you know if there is one for each of the datasets? — Afshin Oroojlooy, Aug 17 '19 at 06:06

SecretAgentMan · Accepted Answer · 2019-08-17T15:21:14.133

Time Series Data Library

The time series data library from Rob Hyndman has hundreds (~648+) of time-series data. It depends on what you call "demand" and what you require as a "sufficiently long period of time."

The entire dataset has been migrated to an R package called tsdl. It is also available on GitHub.

You can find descriptions of the data in the meta data.

Select examples:

"Monthly demand repair parts large/heavy equip. Iowa 1972 – 1979"
"Montly av. residential gas usage Iowa (cubic feet)*100 ’71 – ’79"
"Monthly gasoline demand Ontario gallon millions 1960 – 1975"
"Monthly sales of U.S. houses (thousands) 1965 – 1975"
"Monthly Av. residential electricity usage Iowa city 1971 – 1979"
"Monthly car sales in Quebec 1960-1968"
"Monthly U.S. housing starts (privately owned 1-family) 1965 – 1975"
"Four-weekly totals of beer shipments"
"Annual unit price of lynx pelts (Hudson’s Bay company, Canada) 1857 – 1911"
"Annual yield of straw on Broadbalk field at Rothamsted 1852-1925"
"Basic quarterly iron production in Australia: thousand tonnes. Mar 1956 – Sep 1994"
"Monthly beer production in Australia: megalitres."
"Quarterly beer production in Australia: megalitres. March 1956 – June 1994"

There is another way to see data descriptions:

The raw data is also available. If you drill down, there is a README.txt for each one (e.g. example for "monthly"). These text files are viewable on GitHub in your browser.

Real world demand dataset

1 Answers1

Time Series Data Library

Select examples: