If you run

    from mlxtend.frequent_patterns import apriori
    from mlxtend.frequent_patterns import association_rules
    dataframe = apriori(dataframe, min_support=min_support, use_colnames=True)
    dataframe = association_rules(dataframe, metric="lift", min_threshold=min_threshold)
    dataframe['antecedents_length'] = dataframe['antecedents'].str.len()
    dataframe['consequents_length'] = dataframe['consequents'].str.len()

In pycharm sciView the column name in antecedEnt while in jupyter notebook name is antecedAnt, is the column name specific to platform? Obviously this breaks code. E and A varies.
I ran exactly same code on same dataframe and got column names different

你好，如果我的數據中，每筆訂單同樣商品出現超過1次，那在應用這個套件，他會只算出現一次嗎？

您好，这个仓库很棒，但就是输出格式完全看不懂，ordered_statistics.......这是什么意思啊？

Thanks for the great library! When I was trying to adopt this method on other datasets, such as the census_income from UCI (https://archive.ics.uci.edu/ml/machine-learning-databases/adult/), I found it difficult to set the parameters to get feasible rules. Do you have any advice on this?

Thanks!

Hi,
I am looking to use the apyori package to do some association rule mining on the attached data set. Can you please tell me how to use this package to load the data and work on an ipython notebook? The instructions provided are for command line interface and I am unable to load the data itself.

Can you tell me where am I going wrong?

Regards,
Jash
data_demo.zip

Hi,

Can I please know if the rules returned from the algorithm are sorted by any measure? Or if there's a method to retrieve the top-n most relevant rules?

It seems like the items_add contains only one item. e.g (1,2,5) --> (3,4) will be ignored even if this rule satisfies both minsup and minconf requirements.

Hi!

Thank you by the code, looks like very good!

My point is:
It's normal appear rules with same products on inverse order?

For example:

Base_Item | Append_Item | Support | Confidence | Lift
Rice | Egg | 0.059533 | 0.911369 | 7.44109
Egg | Rice | 0.059533 | 0.486073 | 7.44109

And if it's normal, why this happen? and why the Confidence is different?

Thanks!!!!

Hi everyone,

I have a strange output where the same antecedent with different consequent has different support values. Isn't support only calculated with antecedent frequency on the dataset?

This is my output:

`RelationRecord(items=frozenset({'1 seg_actrab', '0 seg_auto'}), support=0.02740837419309323, ordered_statistics=[OrderedStatistic(items_base=frozenset({'1 seg_actrab'}), items_add=frozenset({'0 seg_auto'}), confidence=0.5803921568627451, lift=3.41041294536262)])

RelationRecord(items=frozenset({'1 seg_actrab', '1 seg_outros'}), support=0.012584218138205934, ordered_statistics=[OrderedStatistic(items_base=frozenset({'1 seg_actrab'}), items_add=frozenset({'1 seg_outros'}), confidence=0.2664799253034547, lift=3.2735085700043927)])`

Hi
After I have found the rules with a minimum support and a minimum confidence, I would sort the rules found by confidence, is it possible?

It is the code that I wrote
rules = list(apriori(dataset, min_support=0.003, min_confidence=0.3, min_lift=1.01))
Ty

Please upload the package to anaconda as well.

Hey there,

I'm trying to understand the output of the Apyori algorithm. When I'm running the algorithm, it appears that the same rules are appearing more than once, with different Support-Confidence & Lift.

Is it normal, and if it is, what does it mean exactly?

From earlier tutorials of apyori that are available online, it looks like there was a rule for a minimum length of items that should be considered. It would be great to have this feature back in place.

My understanding is that would also reduce the computational burden when the ultimate goal is to find the transaction patterns for a minimum number of items.