🐍 一、pd顶层函数

IO（现行）

read_csv/ read_excel/ read_parquet/ read_json/ read_sql/ read_html/ read_clipboard/ read_feather/ read_orc/ read_spss/ read_stata/ read_sas/ read_xml/ read_fwf/ read_pickle

to_pickle/read_iceberg

构造

DataFrame/ Series/ Index/ MultiIndex/ RangeIndex/ Categorical/ CategoricalIndex/ Interval/ IntervalIndex/ Period/ PeriodIndex/ Timedelta/ TimedeltaIndex/ Timestamp/ DatetimeIndex/ TimeDelta/ DateOffset

工具/判断

concat/ merge/ get_dummies/ cut/ qcut/ melt/ pivot/ pivot_table/ crosstab/ wide_to_long/ factorize/ unique/ value_counts

isin/ notna/ isna/ isnull/ notnull

api.types.is_numeric_dtype/ is_string_dtype/ is_categorical_dtype/ is_datetime64_any_dtype/ is_timedelta64_dtype/ is_bool_dtype/ is_integer/ is_float/ is_dict_like/ is_list_like/ pandas.api.extensions.register_extension_dtype

时间

to_datetime/ to_timedelta/ to_numeric/ date_range/ timedelta_range/ period_range/ bdate_range/ infer_freq

offsets.*（Day / BusinessDay / MonthEnd / MonthStart / YearEnd / QuarterEnd / Week / Hour …）

选项

pd.set_option/ pd.get_option/ pd.reset_option/ pd.describe_option

pd.options.mode.copy_on_write（3.0 默认 True，无需再设）

🆕 3.0 新增顶层：pd.col()（延迟列引用，用于 assign/ loc里替代 lambda）

📊 二、DataFrame（现行 public 方法/属性，剔除 removed 后）

构造/查看：__init__/ from_dict/ from_records/ from_spmatrix/ info/ head/ tail/ describe/ memory_usage/ attrs

形状/元数据：shape/ axes/ ndim/ size/ values/ dtypes/ columns/ index/ style/ flags

选择/索引：

__getitem__/ __setitem__/ loc/ iloc/ at/ iat/ get/ lookup

filter/ select_dtypes/ set_index/ reset_index/ reindex/ reindex_like/ rename/ rename_axis/ swaplevel/ set_axis/ take/ xs/ get_indexer/ get_loc

删/改/排：

drop/ drop_duplicates/ dropna/ droplevel

sort_values/ sort_index/ sort_columns

transpose/ T/ swapaxes

astype/ convert_dtypes/ infer_objects（⚠️ copy kwd deprecated）

replace/ fillna/ ffill/ bfill/ interpolate/ where/ mask/ case_when

clip

变形/组合：

melt/ pivot/ pivot_table/ stack/ unstack/ explode/ squeeze/ expand_dims

join/ merge/ append

assign/ eval/ query/ pipe/ transform/ apply/ applymap

分组/窗口：groupby/ rolling/ expanding/ ewm/ resample

统计聚合（每列逐算，也可 axis=0/1 但 axis 在 3.0 很多地方被削）：

sum/ mean/ median/ min/ max/ std/ var/ sem/ prod/ all/ any/ count/ abs/ cumsum/ cumprod/ cummax/ cummin/ cumcount/ diff/ pct_change/ rank/ quantile/ mad/ skew/ kurt/ mode/ nunique/ value_counts/ idxmax/ idxmin/ expanding/ rolling/ ewm

IO 出站：to_csv/ to_excel/ to_parquet/ to_json/ to_sql/ to_html/ to_markdown/ to_clipboard/ to_feather/ to_orc/ to_xml/ to_latex/ to_string/ to_dict/ to_records/ to_numpy/ to_period/ to_timestamp/ to_datetime

🆕 to_iceberg

时序专属：shift/ tshift/ asfreq/ resample/ tz_localize/ tz_convert/ truncate/ first/ last

画图：plot/ hist/ boxplot

其他：compare/ equals/ copy/ between_time/ at_time/ first_valid_index/ last_valid_index/ isna/ notna/ isnull/ notnull/ empty/ pop/ insert/ duplicate→ 应是 duplicated

📈 三、Series（现行）

大部分与 DataFrame 同名方法语义对齐，单列特有：

__init__/ from_array/ info？Series 无 info → 去掉

values/ dtype/ name/ index/ shape/ ndim/ size/ T

iloc/ loc/ at/ iat/ get

cat/ str/ dt/ sparse/ plot/ list→ 无，忽略

append / replace/ fillna/ ffill/ bfill/ interpolate/ where/ mask/ case_when

clip/ astype/ convert_dtypes

map/ apply/ agg/ aggregate/ transform/ pipe

groupby/ rolling/ expanding/ ewm/ resample

value_counts/ unique/ nunique/ factorize/ between/ isin/ isin

duplicated/ drop_duplicates/ dropna/ sort_values/ sort_index

cumsum/ cumprod/ cummax/ cummin/ diff/ pct_change/ shift/ tshift

idxmax/ idxmin/ argmax/ argmin/ argsort/ reorder_levels→ Series 无

explode/ repeat

to_list/ tolist/ to_numpy/ to_dict/ to_frame/ to_xarray/ to_period/ to_timestamp

view/ ravel/ int/ float

🗂️ 四、Index / MultiIndex / RangeIndex

Index现行：astype/ map/ where/ mask/ rename/ set_names/ get_level_values（MI） / swaplevel/ reorder_levels/ droplevel/ append/ difference/ intersection/ union/ symmetric_difference/ is_/ isin/ duplicated/ fillna/ dropna/ sort_values/ sortlevel/ unique/ argmin/ argmax/ min/ max/ any/ all/ to_list/ to_numpy/ to_frame/ to_series

format（→ astype(str)）

MultiIndex加：from_arrays/ from_tuples/ from_product/ get_loc/ get_loc_level/ get_locs/ set_codes/ set_levels

🧮 五、GroupBy 体系（DataFrameGroupBy/ SeriesGroupBy）

agg/ aggregate/ apply/ transform/ pipe

size/ count/ sum/ mean/ median/ std/ var/ sem/ min/ max/ prod/ first/ last/ nth/ n/ ngroup/ cumcount/ rank/ quantile/ pct_change/ resample/ rolling/ expanding/ ewm/ head/ tail/ hist/ boxes→ 无

corr/ cov/ corrwith

groups/ indices/ obj/ grouper

__iter_

🪟 六、Window（rolling/ expanding/ ewm返回对象的方法）

mean/ std/ var/ sum/ min/ max/ count/ median/ quantile/ apply/ agg/ aggregate/ cov/ corr/ skew/ kurt/ sem

rolling特有：center/ win_type/ closed

expanding特有：min_periods

ewm特有：alpha/ span/ halflife/ com/ mean/ std/ var/ corr/ cov

🔤 七、Accessor 三件套（现行）

.str（Series.str 下）:lower/ upper/ title/ capitalize/ swapcase/ casefold/ strip/ lstrip/ rstrip/ pad/ center/ contains/ startswith/ endswith/ find/ rfind/ findall/ match/ fullmatch/ replace/ extract/ extractall/ split/ rsplit/ partition/ rpartition/ cat/ join/ len/ character_count/ unicode_normalize/ normalize/ translate/ wrap/ slice/ slice_replace/ get/ get_dummies/ index/ rindex/ isalnum/ isalpha/ isdigit/ isspace/ islower/ isupper/ istitle/ isnumeric/ isdecimal/ isidentifier/ count/ decode/ encode/ zfill

.dt（Series.dt 下，时间专属）:year/ month/ day/ hour/ minute/ second/ microsecond/ nanosecond/ dayofweek/ day_of_week/ day_of_year/ dayofyear/ quarter/ is_month_start/ is_month_end/ is_year_start/ is_year_end/ is_leap_year/ days_in_month/ date/ time/ tz/ fold/ to_pydatetime/ to_timestamp/ to_period/ strftime/ floor/ ceil/ round/ day_name/ month_name/ total_seconds/ seconds/ microseconds/ na