运行旅行推荐系统代码遇IndexError：索引越界问题求助

问题背景 #

运行智能旅行推荐系统代码时，触发IndexError: indices are out-of-bounds和IndexError: positional indexers are out-of-bounds错误，无法确定是包更新还是循环逻辑问题，相关代码及报错栈如下：

核心代码片段 #

%%capture
final = dict()
final['timeofday'] = []
final['image'] = []
final['name'] = []
final['location'] = []
final['price'] = []
final['rating'] = []
final['category'] = []

for i in range(1,(end_date - begin_date).days+2):
    for j in range(2):
        final['timeofday'].append('Morning')
    for j in range(2):
        final['timeofday'].append('Evening')

for i in range(len(final['timeofday'])): 
    if i%4 == 0: 
        final = top_recc(with_url, final)
    else:
        final = find_closest(with_url, final['location'][-1],final['timeofday'][i], final)

完整报错栈 #

IndexError                                Traceback (most recent call last)
File ~\anaconda3\lib\site-packages\pandas\core\indexing.py:1482, in _iLocIndexer._get_list_axis(self, key, axis)
   1481 try:
-> 1482     return self.obj._take_with_is_copy(key, axis=axis)
   1483 except IndexError as err:
   1484     # re-raise with different error message

File ~\anaconda3\lib\site-packages\pandas\core\generic.py:3716, in NDFrame._take_with_is_copy(self, indices, axis)
   3709 """
   3710 Internal version of the `take` method that sets the `_is_copy`
   3711 attribute to keep track of the parent dataframe (using in indexing
   (...)
   3714 See the docstring of `take` for full explanation of the parameters.
   3715 """
-> 3716 result = self.take(indices=indices, axis=axis)
   3717 # Maybe set copy if we didn't actually change the index.

File ~\anaconda3\lib\site-packages\pandas\core\generic.py:3703, in NDFrame.take(self, indices, axis, is_copy, **kwargs)
   3701 self._consolidate_inplace()
-> 3703 new_data = self._mgr.take(
   3704     indices, axis=self._get_block_manager_axis(axis), verify=True
   3705 )
   3706 return self._constructor(new_data).__finalize__(self, method="take")

File ~\anaconda3\lib\site-packages\pandas\core\internals\managers.py:897, in BaseBlockManager.take(self, indexer, axis, verify)
    896 n = self.shape[axis]
-> 897 indexer = maybe_convert_indices(indexer, n, verify=verify)
    899 new_labels = self.axes[axis].take(indexer)

File ~\anaconda3\lib\site-packages\pandas\core\indexers\utils.py:292, in maybe_convert_indices(indices, n, verify)
    291     if mask.any():
-> 292         raise IndexError("indices are out-of-bounds")
    293 return indices

IndexError: indices are out-of-bounds

The above exception was the direct cause of the following exception:

IndexError                                Traceback (most recent call last)
Input In [8], in <cell line: 16>()
     16 for i in range(len(final['timeofday'])): 
     17     if i%4 == 0: 
---> 18         final = top_recc(with_url, final)
     19     else:
     20         final = find_closest(with_url, final['location'][-1],final['timeofday'][i], final)

File ~\Desktop\Intelligent-Travel-Recommendation-System-master\attractions_recc.py:114, in top_recc(with_url, final)
    112 i=0
    113 while(1):
-> 114     first_recc = with_url.iloc[[i]]
    115     if(first_recc['name'].values.T[0] not in final['name']):
    116         final['name'].append(first_recc['name'].values.T[0])

File ~\anaconda3\lib\site-packages\pandas\core\indexing.py:967, in _LocationIndexer.__getitem__(self, key)
    964 axis = self.axis or 0
    966 maybe_callable = com.apply_if_callable(key, self.obj)
-> 967 return self._getitem_axis(maybe_callable, axis=axis)

File ~\anaconda3\lib\site-packages\pandas\core\indexing.py:1511, in _iLocIndexer._getitem_axis(self, key, axis)
   1509 # a list of integers
   1510 elif is_list_like_indexer(key):
-> 1511     return self._get_list_axis(key, axis=axis)
   1513 # a single integer
   1514 else:
   1515     key = item_from_zerodim(key)

File ~\anaconda3\lib\site-packages\pandas\core\indexing.py:1485, in _iLocIndexer._get_list_axis(self, key, axis)
   1482     return self.obj._take_with_is_copy(key, axis=axis)
   1483 except IndexError as err:
   1484     # re-raise with different error message
-> 1485     raise IndexError("positional indexers are out-of-bounds") from err

IndexError: positional indexers are out-of-bounds

报错原因分析 #

1. 核心触发点：错误来自top_recc函数的第114行with_url.iloc[[i]]——当变量i的取值超过with_url这个DataFrame的最大行索引时，就会触发索引越界。
2. 逻辑缺陷：
   • top_recc里的while(1)是无限循环，没有设置终止条件，每次循环i递增后最终会超出with_url的行数上限。
   • 外层循环的次数由len(final['timeofday'])决定，这个值根据旅行天数生成（每天4个时段），但如果旅行天数对应的推荐需求超过了with_url里的景点总数，就会导致反复调用top_recc时耗尽所有可推荐景点，最终触发索引越界。
3. 排除包更新因素：pandas中iloc的索引逻辑是基础行为，版本更新不会改变“索引超出范围报错”的规则，因此不是包更新导致的问题。

解决方案 #

1. 修复top_recc函数的循环逻辑 #

给无限循环加上终止条件，避免索引越界：

def top_recc(with_url, final):
    i = 0
    max_valid_index = len(with_url) - 1  # 获取DataFrame的最大合法索引
    while i <= max_valid_index:
        first_recc = with_url.iloc[[i]]
        target_name = first_recc['name'].values.T[0]
        if target_name not in final['name']:
            # 补充其他字段的赋值逻辑
            final['name'].append(target_name)
            final['image'].append(first_recc['image'].values.T[0])
            final['location'].append(first_recc['location'].values.T[0])
            final['price'].append(first_recc['price'].values.T[0])
            final['rating'].append(first_recc['rating'].values.T[0])
            final['category'].append(first_recc['category'].values.T[0])
            return final
        i += 1
    # 所有景点已推荐完毕，直接返回原final
    return final

2. 限制外层循环的执行次数 #

在外层循环中加入检查，当没有更多可推荐的景点时，提前终止循环：

max_available_recc = len(with_url)  # 可推荐的最大景点数
for i in range(len(final['timeofday'])): 
    # 若已推荐完所有景点，直接终止循环
    if len(final['name']) >= max_available_recc:
        break
    if i%4 == 0: 
        final = top_recc(with_url, final)
    else:
        final = find_closest(with_url, final['location'][-1],final['timeofday'][i], final)

3. 可选优化：提升重复检查效率 #

如果final['name']的长度很大，target_name not in final['name']的效率很低，可以改用集合来记录已推荐的景点：

# 初始化时添加集合存储已推荐名称
final['recommended_names'] = set()

# 在top_recc里修改检查逻辑
if target_name not in final['recommended_names']:
    final['name'].append(target_name)
    final['recommended_names'].add(target_name)
    # 其他字段赋值...

内容的提问来源于stack exchange，提问作者Mahdi Karblaee