0

我有以下数据集：

dataset = [[0, 'Milk', 'Onion', 'Nutmeg', 'Kidney Beans', 'Eggs', 'Yogurt'],
           [0, 'Dill', 'none', 'Nutmeg', 'Kidney Beans', 'Eggs', 'Eggs'],
           [0,'Dill', 'milk'],
           [1,'Dill'],
           [1,'Corn', 'Onion', 'Onion', 'Kidney Beans', 'Ice cream', 'Eggs'],
           ['......','........','........','........','........','........','........',],
          
           [24,'Corn', 'Onion', 'Onion', 'Kidney Beans', 'Ice cream', 'Eggs']]

df = pd.DataFrame(dataset)

    0   1         2       3        4             5            6
0   0   Milk    Onion   Nutmeg  Kidney Beans    Eggs        Yogurt
1   0   Dill    none    Nutmeg  Kidney Beans    Eggs         Eggs
2   0   Dill    milk    None    None            None         None
3   1   Dill    None    None    None            None         None
4   1   Corn    Onion   Onion   Kidney Beans    Ice cream    Eggs
5 ......    ........    .....   ........      ........     .......
6   24  Corn    Onion   Onion   Kidney Beans    Ice cream    Eggs

适合关联规则学习算法

from mlxtend.frequent_patterns import fpgrowth
frequent_itemsets_fp=fpgrowth(df, min_support=0.001, use_colnames=True)

from mlxtend.frequent_patterns import association_rules
rules_fp = association_rules(frequent_itemsets_fp, metric="lift").sort_values ("lift", ascending=True).reset_index(drop=True

列0指定这些事务发生的时间。小时有一天的持续时间-->0-23 hour

我想要的是使用每次作为输入对应于一小时的行来训练算法

所以首先我想使用O小时的所有行训练算法并保存结果，然后是小时的行，直到1小时结束。

有任何想法吗？

4

0 回答 0