如果值有新数据，则为dataframe迭代列填充

Question

我一直在研究熊猫数据帧。陷入某种逻辑。有人能帮上忙吗：

这是我的输入数据框：

**df4**

Data1   Description Associated Data
F104    TO  F5334
F104    TO  F105
F105    TO  F1212
F105    TO  F5332
F105    IS  F104
F5332   IS  F105
F5334   IS  F104
F1000   IS  F1212
F1015   IS  F1242
F1015B  IS  F1242B
F1015B  IS  F1242
F1212   TO  F1000
F1212   TO  F1242B
F1212   IS  F105
F1242   TO  F1015
F1242   TO  F1015B
F1242B  TO  F1015B
F1242B  IS  F1212

这里有3列： 数据1（键）、描述（用作条件）、关联数据（值）

预期产出：

F104    F5334       
F104    F105    F1212 
F104    F105    F5332   
F104    F105    F1212 F1000 
F104    F105    F1212 F1242B
...............

我的代码：

df6 = pd.DataFrame(columns=["Is_Upgrade"])
df4 = df.copy()
for i in range(len(df4)):
    print(i)
    bas_part = df4.loc[i]["Data1"]
    type_part = df4.loc[i]["Description"]
    #associate = df4.loc[i]["Associated Data"]
    if((str(type_part) == "TO") == True):
        df6.loc[i,"Data1"]= str(bas_part)
        df6.loc[i,"Upgrade_use"] = df4.loc[i]["Associated Data"]
    elif((str(type_part) == "IS") == True):
        #if(df6.loc[i]["Upgrade"] == ""):
            df6.loc[df6["Data1"]== str(bas_part),"Is_Upgrade"] = df4.loc[i]["Associated Data"]

取得的结果：

F104    F105    F1212
F104    F105    F5332
F105    F1212   F1000
F105    F1212   F1242B
nan     F104    F5334
nan     F104    F105
F1212   F1242B  F1015B
nan     F1242   F1015
nan     F1242   F1015B

详细说明：

如果描述列值为“TO”，则将关联数据添加到实际数据的右侧

迭代1：

F104, F5334  

迭代2：因为我们有条件再次创建具有相同值的新行

F104 F5334
F104 F105

迭代3：因为我们有条件再次创建具有相同值的新行

F104 F5334
F104 F105 F1212

迭代4：因为我们有条件再次创建具有相同值的新行

F104 F5334
F104 F105 F1212
F104 F105 F5332

迭代5：由于我们有条件“是”，现在检查data1左边的值是否与associate相同如果“是”通过，否则添加新值将现有数据移动到新列

F104 F5334
F104 F105 F1212
F104 F105 F5332

迭代7：由于我们有条件“是”，现在检查data1左边的值是否与associate相同如果“是”通过，否则添加新值将现有数据移动到新列

F104 F5334
F104 F105 F1212
F104 F105 F5332

等等

有人能帮忙吗

Answer 1

方法是创建一个函数，用于检查数据帧中的索引值：

def right_tree_pn(df5,base_partnum):
 if(len(df5) == 0):
            parts_list = []
 else:
            new_pn_list = df5
            parts_list = new_pn_list.replace("[","").replace("]","").replace("'","").split(", ")
 if(base_partnum in parts_list):
     base_index = "Yes"
 else :
     base_index = "No"
 
 return base_index

现在创建一个for循环：

for i in range(len(df4)):
    print(i)
    base = df4.loc[i]["Data1"]
    type_part = df4.loc[i]["Description"]
    associate_data = df4.loc[i]["Associated Data"]
    test_parts2 = base + "," + associate_data
    test_parts =test_parts2.split(",")
    #check for condition
    if((str(type_part) == "TO") == True):
    #initially the data frame is empty
        if(len(df5) == 0):
           parts_list = []
     ## Check for value . it is 0 for first step
           if(len(parts_list) > 0):
                index = parts_list.index(base)
                #index_val = len(df5)
           else:
                index = 0
                index_val = len(df5)
               # Insert to first Position as index i.e base value
                parts_list.insert(index, base)
           if(associate_data in parts_list):
                associate_index = parts_list.index(associate_data)
           else:
                associate_index = index
        #########################################
           if((associate_index < index) or (associate_index > index) == True):
            #df5.loc[new_index,"Data"] =  str(parts_list)
                pass
           else:
                #add Associate data to next index value i.1 1 for first step
                 parts_list.insert(index + 1,associate_data)
           df5.loc[index_val ,"Data"] = str(parts_list)
           df5.loc[index_val ,"Base_PN"] = str(parts_list[0])
   ######### Add to new line if dataframe is not empty 
        else:
            df_index = []
    ###############Get the index values from the dataframe if the base data exists
            for x in range(len(df5)):
                df6 = right_tree_pn(df5.loc[x]["Data"],base)
                if(str(df6) == "Yes"):
                    df_index1 = x
                    df_index.append(df_index1)
#Check length of the index if it is 1 perform following operations
            if(len(df_index) == 1):
                df_index = df_index[0]
                new_pn_list = df5.loc[df_index]["Data"]
                parts_list1 = new_pn_list.replace("[","").replace("]","").replace("'","").replace(" ","").split(",")
                ######List compare
                if(all(w in parts_list1 for w in test_parts) == True):
                    pass
                else:
                    index = parts_list1.index(base)
                    if((index == 0 and len(parts_list1) != 0)) == True:
                        ass_pasrt2 = parts_list1[index + 1]
                        if(index == 0 and associate_data != ass_pasrt2) == True:
                            parts_list = []
                            index_val = len(df5)
                        else:
                             parts_list = parts_list1
                             index_val = df_index
                    elif(((index == 0 and len(parts_list1) == 0)) ==True):
                            parts_list = []
                            index_val = len(df5)
                    else:
                        index2 = len(parts_list1) - 1
                        if(index2 != index):
                            ass_pasrt2 = parts_list1[index + 1]
                        else:
                            ass_pasrt2 = ""
                        if((associate_data == ass_pasrt2)== True):
                            parts_list = parts_list1
                            index_val = df_index
                        elif((ass_pasrt2 == "") == True):
                            parts_list = parts_list1
                            index_val = df_index
                        else:
                            parts_list = parts_list1
                            if(ass_pasrt2 in parts_list):
                                parts_list.remove(ass_pasrt2)
                            index_val = len(df5)
                    if(len(parts_list) > 0):
                        index = parts_list.index(base)
                    else:
                            index = 0
                            parts_list.insert(index, base)
                    if(associate_data in parts_list):
                        associate_index = parts_list.index(associate_data)
                    else:
                        associate_index = index
        #########################################
                    if((associate_index < index) or (associate_index > index) == True):
            #df5.loc[new_index,"Data"] =  str(parts_list)
                        pass
                    else:
                        parts_list.insert(index + 1,associate_data) 
                        parts_list2 = list(OrderedDict.fromkeys(parts_list))
                    df5.loc[index_val,"Data"] = str(parts_list2)
                    df5.loc[index_val ,"Base_PN"] = str(parts_list2[0])
  #### If the base is not present in the data frame create new row similar to first iteration
            elif(len(df_index) == 0):
                parts_list = []
                if(len(parts_list) > 0):
                    index = parts_list.index(base)
                    #index_val = len(df5)
                else:
                    index = 0
                    parts_list.insert(index, base)
                    index_val = len(df5)
                if(associate_data in parts_list):
                    associate_index = parts_list.index(associate_data)
                else:
                    associate_index = index
        #########################################
                if((associate_index < index) or (associate_index > index) == True):
            #df5.loc[new_index,"Data"] =  str(parts_list)
                    pass
                else:
                    parts_list.insert(index + 1,associate_data)
                df5.loc[index_val ,"Data"] = str(parts_list)
                df5.loc[index_val ,"Base_PN"] = str(base)
            else:
#Code for multiple index data 
                for z in range(len(df_index)):
                    index_value = df_index[z]
                    new_pn_list = df5.loc[index_value]["Data"]
                    parts_list1 = new_pn_list.replace("[","").replace("]","").replace("'","").split(", ")
                    if(all(w in parts_list1 for w in test_parts) == True):
                        pass
                    else:
                        index = parts_list1.index(base)
                        if((index == 0 and len(parts_list1) != 0)) ==True:
                            parts_list = parts_list1
                        elif(((index == 0 and len(parts_list1) == 0)) ==True):
                            parts_list = []
                        else:
                            parts_list = parts_list1
                    #    parts_list = parts_list1
                    if(len(parts_list) > 0):
                        index = parts_list.index(base)
                    else:
                            index = 0
                            parts_list.insert(index, base)
                    if(associate_data in parts_list):
                        associate_index = parts_list.index(associate_data)
                    else:
                        associate_index = index
        #########################################
                    if((associate_index < index) or (associate_index > index) == True):
            #df5.loc[new_index,"Data"] =  str(parts_list)
                        pass
                    else:
                        parts_list.insert(index + 1,associate_data)
                        parts_list2 = list(OrderedDict.fromkeys(parts_list))
                    df5.loc[index_value ,"Data"] = str(parts_list2)
                    df5.loc[index_value ,"Base_PN"] = str(parts_list2[0])

创建数据框后，我们必须将数据拆分为多列：

df5 = df5[["Base_PN","Data"]]                   
df5["Data2"] = df5["Data"].str.replace("[","").str.replace("]","").str.replace("'","").str.replace(" ","")
df6 = df5.Data2.apply(lambda x: pd.Series(str(x).split(","))).add_prefix('Level_')

输出：

Level_0 Level_1 Level_2 Level_3 Level_4 
                    
F104    F5334       
F104    F105    F1212 
F104    F105    F5332   
F104    F105    F1212 F1000 
F104    F105    F1212 F1242B    F1015B