I am looking for a typical sales data set, which I can use for a time series prediction. I especially want to show how to decompose the seasonal component.
A good example would be a fictional company, or a restaurant that sells much higher volumes in the summer/ or winter.
The US Census Bureau provides a good data source:
U.S. Census Bureau's Monthly & Annual Retail Trade
I work at Quandl and here are some databases you might want to check out:
US Census Bureau (free) - Here are the datasets you'll see if you search for "restaurant sales" within this database:
https://www.quandl.com/data/USCENSUS-U-S-Census-Bureau?keyword=restaurant%20sales
Sales Surprises - This is a premium database with sales data for over 5000 companies, including specific restaurants.
https://www.quandl.com/data/ZSS-Sales-Surprises
(Full list of companies covered are listed here: https://www.quandl.com/api/v3/databases/ZSS/codes)
Hope this helps...
Sales volumes are difficult to find, explicitly broken out would be difficult, however if you look at staples such as soft drink distributors or restaurant's quarterly reports, you can get a basic break out. The first two to look at would be Coca Cola and Pepsi:
Looking on YCharts, this gives you a quick view of seasonality, but you only get ten page views before you must register.
Publicly traded restaurants can give you an idea of additional seasonal variations:
Electronics are difficult as there is such variation
You can find the open source versions from their respective Quarterly Earnings Reports:
The Monash Time Series Forecasting Repository contains multiple different time series datasets from various domains. These have been collected from forecasting competitions or other previous forecasting use, so you should also be able to learn something from previous work that used a particular dataset.
Some of the datasets included should show clear seasonality, even if they are not sales. Specifically, I am thinking of the Tourism and the Electricity datasets. These may even exhibit multiple seasonalities - for instance, the Electricity dataset contains hourly data, and of course there are seasonalities within each day, but the intra-daily seasonality differs between weekdays and weekends. (For that matter, so does StackExchange traffic, which you can also download.)
An introduction and description of the repository, along with more information about the datasets, is given by Godahewa et al. (2021).
Alternatively, take a look at the M5 forecasting competition, which I believe is not (yet) contained in the Monash Repository. It's daily sales on a store $\times$ SKU level from multiple Walmart stores. The weekly seasonality may not be visible on the lowest level of aggregation, but it should become clear once you aggregate SKUs up to categories.