Part two of the prognostica Winter Challenge is online!

Have you missed part one? Click here to see what this is all about!

Part two of the prognostica Winter Challenge: Outliers in order behavior

In planning processes it is often helpful to know what the smallest packaging is in which a product is sold. It can be assumed that eggs are always sold in multiples of 6, as eggs are sold in cartons of 6. However, if we look at the actual sales figures, we see that this assumption is not correct. This is because eggs are also sold in cartons of 10. Such phenomena are to be tracked down in this task.

Such additional information is often available in the industry. Unfortunately, this information is often not entirely reliable. The following task is about recognizing whether the specified packaging size is a reliable indicator.

We have prepared a few time series for this. You can find them here. You can find the size of the package in the file names (e.g. the file file_2_quantity_86.csv contains a time series with a package size of 86. file_2 simply means that it is the second time series). For this task, we assume that there is only exactly one package size per time series. Your code should read in a file and identify whether the following rules are observed:

1. A maximum of 7 values are NOT multiples of the specified package size
2. Never more than 3 consecutive values are NOT multiples of the package size

To do this, write a function check_quantity_validity(filepath) in Python or R, which receives the file name as input and outputs the result in a JSON in the following format:

{"is_valid": "True"}

oder

{"is_valid": "False"}

If you want to present us with additional results, simply store them as another field in the JSON object. Show us how to call your function in a Jupyter notebook. We are looking forward to testing your solution with more examples. Send us your submission to winter@challenge.prognostica.de. We are looking forward to it.