I want to use association rules with times of the day (in categories) and products. For example, product A is more often ordered in the morning, whereas product B is more often ordered in the evening.
I have made already two columns which show the time (in categories) and all the products in the specific order. When I combine the columns, I experience some weird formatting with brackets or quotation marks in between. This association rule learner still produces an output but sometimes there are rules such as: ["productA --> evening] and [productA --> "evening] . So, the rule is in fact the same but Knime does not recognise it as such.
I attached a snip from my workflow. Maybe I have to use an entirely different approach?
Thank you all!
are you sure those quotes do not come from your input data? Maybe it is not cleaned properly? You can check that using the Duplicate Row Filter and a subsequent Sorter node.
Yes, thank you I have checked it.
How would you approach the following then? Based on the time of the day (in categories such as morning, noon, evening…) and an itemset, I want to use association rules or any other type of clustering. Basically, if certain products are ordered at specific times.
I have orders with different order lines (this is why I used the groupBy delivery node).
Would you use a very different approach?
can’t you just group by time and product ID and then count the occurrence? You can then compare that to the occurrence of the product ID over all times and calculate the fraction for which it it occurs at each time.