使用 Surprise 库时出错，说我的项目不是训练集的一部分

如何解决使用 Surprise 库时出错，说我的项目不是训练集的一部分

我正在按照 Surprise 库的文档使用协同过滤构建推荐引擎。

我像这样创建了内部和外部 ID

# create ID for each item
# dfsurprise is the name of the dataframe that holds my data (user,item,rating)

grouped_name = dfsurprise.groupby('item')

temp_df = grouped_name.count()
temp_df_idx = pd.DataFrame(temp_df.index)


temp_df_idx['itemID'] = temp_df_idx.index
dict_df=temp_df_idx[['itemID','item']]


desc_dict = dict_df.set_index('item').to_dict()
new_dict = desc_dict['itemID']

dfsurprise['itemID'] = dfsurprise.item.map(new_dict)



# create userID for each user
grouped_user = dfsurprise.groupby('user')

temp_df_user = grouped_user.count()
temp_df_user_idx = pd.DataFrame(temp_df_user.index)


temp_df_user_idx['userID']=temp_df_user_idx.index
dict_df_user=temp_df_user_idx[['userID','user']] 


desc_dict_user = dict_df_user.set_index('user').to_dict()
new_dict_user = desc_dict_user['userID']

dfsurprise['userID'] = dfsurprise.user.map(new_dict_user)

然后我有一个函数可以将 id 转换为项目，反之亦然。

def read_item_names():
    """
    return two mappings to convert raw ids into item names 
    and item names into raw ids.
    """

    file_name = dict_df
    rid_to_name = {}
    name_to_rid = {}

    unique_items = len(dfsurprise.item.unique())
    
    for i in range(unique_items):  
        line = file_name.iloc[i]
        rid_to_name[line[0]] = line[1]
        name_to_rid[line[1]] = line[0]

    return rid_to_name,name_to_rid

最后，我有一个使用 K 最近邻算法输出推荐的函数。这是我收到错误的地方：

def get_rec(item_name,k_):
    """
    Input item name and returns k recommendations
    based on item similarity
    
    Input: String,integer
    Output: String
    """
    output = []
    
    item = str(item_name)

    # Read the mappings raw id <-> item name
    rid_to_name,name_to_rid = read_item_names()

    # Retrieve inner id of the item
    item_input_raw_id = name_to_rid[item]
    item_input_inner_id = algo.trainset.to_inner_iid(item_input_raw_id) #ERROR

    
    K = k_
    
    # Retrieve inner ids of the nearest neighbors of the item
    item_input_neighbors = algo.get_neighbors(item_input_inner_id,k=K)

    
    # Convert inner ids of the neighbors into names.
    item_input_neighbors = (algo.trainset.to_raw_iid(inner_id)
                                for inner_id in item_input_neighbors)
    item_input_neighbors = (rid_to_name[rid]
                                for rid in item_input_neighbors)

    for item_ in item_input_neighbors:
        output.append(item_)
        
    return output

然后我训练算法并调用 get_rec 函数：

# Train the algortihm to compute the similarities between items (item-item collaborative filtering)
reader = Reader(rating_scale=(1,5))
data = Dataset.load_from_df(dfsurprise[['user','item','rating']],reader)
trainset = data.build_full_trainset()
sim_options = {'name': 'pearson_baseline','user_based': False}
algo = sp.KNNBaseline(sim_options=sim_options)
algo.fit(trainset)

在 get_rec 函数中，我得到错误：

ValueError: Item 348 is not part of the trainset.

这是由该行产生的：

item_input_inner_id = algo.trainset.to_inner_iid(item_input_raw_id)

为什么我的物品不属于火车组？