Apriori关联规则算法

实现购物篮分析

# 这里是吧每一个数据加上一个','
df1['Goods'] = df1['Goods'].apply(lambda x:','+x) 
# 这里是将所有买的物品的字符串合在一起
df4=df1.groupby('id').sum()
# 把分好的物品变为列表
df4["Goods"]=df4["Goods"].apply(lambda x:x[1:].split(","))

上面的操作就是为了实现下面的 表

这下面写的是 相关性分析的函数

data=list(df4.Goods)

import copy
def get_all_list(datas):
    all_list=[]
    for data in datas:
        for i in data:
            if i not in all_list:
                all_list.append(i)
    return all_list
def get_C1(all_list,datas,minSupport):
    lenght=len(all_list)
    dict_count={}
    for i in all_list:
        count=0
        for data in datas:
            if i in data:
                count+=1
        dict_count[i]=round(count/lenght,2)
    re_list=[]
    for item in dict_count.items():
        if item[1]>minSupport:
            re_list.append(item[0])
    return re_list
def isin(one,two):
    for i in one:
        if i not in two:
            return False
    else:
        return True
def get_C2(re_list,datas,minSupport):
    lenght=len(datas)
    a=1
    all_list=[]
    for i in re_list:
        for j in re_list[a:]:
            all_list.append([i,j])
        a+=1
    dict_count={}
    for i in all_list:
        count=0
        for data in datas:
            if isin(i,data):
                count+=1
        dict_count[str(i)]=round(count/lenght,2)
    dict_count2=copy.deepcopy(dict_count)
    for item in dict_count2.items():
        if item[1]<minSupport:
            del dict_count[str(item[0])]
    return dict_count
def get_datas(datas,dict_count,contract):
    final_list=[]
    for item in dict_count.items():
        for it in [eval(item[0]),eval(item[0])[::-1]]:
            need=support(datas,it)
            flag=1
            if need<contract:
                flag=0
            final_list.append([it,item[1],need,flag])
    return final_list

def support(datas,item):
    count1=0
    count2=0
    for i in datas:
        if item[0] in i: 
            count1+=1
        if item[1] in i:
            count2+=1
    return round(count1/count2,2)

# 这里传入的data 是 data=list(df4.Goods)
all_list=get_all_list(data)
re_list=get_C1(all_list,data,0.04)
C2=get_C2(re_list,data,0.04)
need_data=get_datas(data,C2,0.35)

table=pd.DataFrame(need_data,columns=["关系","支持度","置信度","是否强关联"])
table["关系"]=table["关系"].apply(lambda x:x[0]+"-->"+x[1])