I have the data of all transactions of the last 5 years from the cafeteria where we eat students.
I have already prepared the data to the extent that I have counted the number of transactions in a 15 minute window and have it in a JSON format.
Also I have an upper and lower limit of the values, respectively I have already scaled the data as well.
I found the following tutorial and was able to customize it with my data: https://jinglescode.github.io/time-series-forecasting-tensorflowjs/
My data is in the following format:
[
{"2021-09-09T08:15:00.000Z":0.0013333333333333333},
...
]
What is the best way to get a realistic forecast? That there is no magic trained model is clear to me.
The data for the last year looks like:
