Apriori | Python Implementation of Apriori Algorithm | Data Mining library

In a dataframe with repeat values, I want the rows for the most frequent n cases, say the two most frequently occurring. The code below does this, selecting rows where x==3 or x==4 and return the rows in that order.

I don't want to have to use the value 5; however, I want some way of programmatically stating the top 2 most frequent x values, without knowing a threshold (5 in this example) apriori. In addition, I would like to order the resulting dataframe by frequency of occurrence, so x==4 rows come before x==3 rows.

I am presuming it is related to count, top_n or slice_max and arrange but maybe not!

Any hints on how to do this with dplyr would be greatly appreciated.

I have a project that I want to find the rules of correlation between goods in the shopping cart .to do this ,I use the ML service(Python) in Sql Server ,and I use the mlxtend library to find the association rule.but the problem I have is that the fpgrowth function apparently uses a lot of memory, to the point where it stops working and gives errors.as far as possible, do the data preprocessing with sql server to be more efficient.

Part of the code :

I am scraping some data from the sec archives. Each xml document has the basic form:

I have a datetime column in a Dataframeconsisting of all the days throughout the year. Every time a date occurs, I would like to count it and store the occurrence and the date. If a date is not occurring: store a zero at the given date, with the date as key.

I have tried by looping over the column and storing the values, for each date, in a dict but could not implement the storing of zero of said date, if it is not occurring.

Example/sample of the column:

I have 12 millions of data from an eshop. I would like to compute association rules using efficient_apriori package. The problem is that 12 millions observations are too many, so the computation tooks too much time. Is there a way how to speed up the algorithm? I am thinking about some Parallel-processing or compile python code into C. I tried PYPY, but PYPY does not support pandas package. Thank you for any help or idea.

If you want to see my code:

I am trying to transform this long dataframe to the wide dataframe with the following logic, the numbering of the columns is not important, what is important that the format stays this way as then I would need to use it for apriori algorithm.

How should a txt or csv file be setup for uploading to weka in order to use apriori? I have tried setting it up as a binary but the associations don't seem to come out correctly. Assuming my database transactions are simple like below what would be the correct way to create a csv or txt file for uploading to weka? The first column is the transaction id and the latter is the items for that transaction.

1 --- {M,O,N,K,E,Y}
2 --- {D,O,N,K,E,Y}
3 --- {M,A,K,E}
4 --- {C,O,O,K,I,E}
5 --- {D,O,O,D,L,E}

I have problem with finding "correct" associations within production data.

The data looks like this

I want to collapse some specific rows of a data.frame (preferably using dplyr in ). Collapsing should aggregate some columns by the functions sum(), others by mean().

As an example, let's add a unique character-based ID to the iris dataset.