Eclat.py

# Eclat Algorithm tutorial from Machine Learning A-Z - SuperDataScience -> Input by Ryan L Buchanan 12OCT20

!pip install apyori

# Import libraries
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd

# Data Preprocessing 
dataset = pd.read_csv('Market_Basket_Optimisation.csv', header = None)
transactions = []
for i in range(0, 7501):
    transactions.append([str(dataset.values[i, j]) for j in range(0, 20)])

# Train the Eclat model on the dataset
from apyori import apriori
rules = apriori(transactions, min_support = 0.003, min_confidence = 0.2, min_lift = 3, min_length = 2,
                max_length = 2)

# Visualze the results
results = list(rules)
print(results)

# Put the results into a well-organized Pandas DataFrame
def inspect(results):
    lhs = [tuple(result[2][0][0])[0] for result in results]
    rhs = [tuple(result[2][0][1])[0] for result in results]
    supports = [result[1] for result in results]
    return list(zip(lhs, rhs, supports))
resultsinDataFrame = pd.DataFrame(inspect(results), columns = ['Product 1', 'Product 2', 'Support'])
print(resultsinDataFrame)

# Display the results sorted by descending lifts
resultsinDataFrame.nlargest(n = 10, columns = 'Support')