is it correct code: cars = generateCARs(txns_train , maxlen= 3, support= 0.1 , confidence = 0.2 )

[Sandy4321]

1
is it correct code to set key parameters ?
2
by the way is it possible to set max number of rules?
or some recommendations how to gt less rules for the approximately same confusion matrix (or F1)
3
what to do with unbalanced data?
4
did you compared performance with with corels ?
https://github.com/fingoldin/pycorels

5
how to set better accuracy by worse recall
or
vice versa ?
6
can predict_proba be used?

---

from pyarc import TransactionDB

from pyarc.algorithms import (

top_rules,

createCARs,

M1Algorithm,

M2Algorithm,

generateCARs

)

import pandas as pd

import numpy as np

data_train = pd.read_csv("iris.csv")

data_test = pd.read_csv("iris.csv")

txns_train = TransactionDB.from_DataFrame(data_train)

txns_test = TransactionDB.from_DataFrame(data_test)

get the best association rules

rules = top_rules(txns_train.string_representation)

convert them to class association rules

#cars = createCARs(rules)

cars = generateCARs(txns_train , maxlen= 3, support= 0.1 , confidence = 0.2 )

classifier = M1Algorithm(cars, txns_train).build()

classifier = M2Algorithm(cars, txns_train).build()

accuracy = classifier.test_transactions(txns_test)

predicted_txns_train = classifier.predict_all(txns_train)

data_train['class'].values

from sklearn.metrics import confusion_matrix

print(confusion_matrix(predicted_txns_train,data_train['class'].values ))

for each_rule in classifier.rules :

print(each_rule)

q=0