zoukankan      html  css  js  c++  java
  • DataFrame基本函数

    属性和数据
    DataFrame.axes                                #index: 行标签;columns: 列标签
    DataFrame.as_matrix([columns])                #转换为矩阵
    DataFrame.dtypes                              #返回数据的类型
    DataFrame.ftypes                              #返回每一列的 数据类型float64:dense
    DataFrame.get_dtype_counts()                  #返回数据框数据类型的个数
    DataFrame.get_ftype_counts()                  #返回数据框数据类型float64:dense的个数
    DataFrame.select_dtypes([include, include])   #根据数据类型选取子数据框
    DataFrame.values                              #Numpy的展示方式
    DataFrame.axes                                #返回横纵坐标的标签名
    DataFrame.ndim                                #返回数据框的纬度
    DataFrame.size                                #返回数据框元素的个数
    DataFrame.shape                               #返回数据框的形状
    DataFrame.memory_usage()                      #每一列的存储

    类型转换
    DataFrame.astype(dtype[, copy, errors])       #转换数据类型
    DataFrame.copy([deep])                        #deep深度复制数据
    DataFrame.isnull()                            #以布尔的方式返回空值
    DataFrame.notnull()                           #以布尔的方式返回非空值

    索引和迭代
    DataFrame.head([n])                           #返回前n行数据
    DataFrame.at                                  #快速标签常量访问器
    DataFrame.iat                                 #快速整型常量访问器
    DataFrame.loc                                 #标签定位,使用名称
    DataFrame.iloc                                #整型定位,使用数字
    DataFrame.insert(loc, column, value)          #在特殊地点loc[数字]插入column[列名]某列数据
    DataFrame.iter()                              #Iterate over infor axis
    DataFrame.iteritems()                         #返回列名和序列的迭代器
    DataFrame.iterrows()                          #返回索引和序列的迭代器
    DataFrame.itertuples([index, name])           #Iterate over DataFrame rows as namedtuples, with index value as first element of the tuple.
    DataFrame.lookup(row_labels, col_labels)      #Label-based “fancy indexing” function for DataFrame.
    DataFrame.pop(item)                           #返回删除的项目
    DataFrame.tail([n])                           #返回最后n行
    DataFrame.xs(key[, axis, level, drop_level])  #Returns a cross-section (row(s) or column(s)) from the Series/DataFrame.
    DataFrame.isin(values)                        #是否包含数据框中的元素
    DataFrame.where(cond[, other, inplace, …])    #条件筛选
    DataFrame.mask(cond[, other, inplace, …])     #Return an object of same shape as self and whose corresponding entries are from self where cond is False and otherwise are from other.
    DataFrame.query(expr[, inplace])              #Query the columns of a frame with a boolean expression.

    二元运算
    DataFrame.add(other[,axis,fill_value])        #加法,元素指向
    DataFrame.sub(other[,axis,fill_value])        #减法,元素指向
    DataFrame.mul(other[, axis,fill_value])       #乘法,元素指向
    DataFrame.div(other[, axis,fill_value])       #小数除法,元素指向
    DataFrame.truediv(other[, axis, level, …])    #真除法,元素指向
    DataFrame.floordiv(other[, axis, level, …])   #向下取整除法,元素指向
    DataFrame.mod(other[, axis,fill_value])       #模运算,元素指向
    DataFrame.pow(other[, axis,fill_value])       #幂运算,元素指向
    DataFrame.radd(other[, axis,fill_value])      #右侧加法,元素指向
    DataFrame.rsub(other[, axis,fill_value])      #右侧减法,元素指向
    DataFrame.rmul(other[, axis,fill_value])      #右侧乘法,元素指向
    DataFrame.rdiv(other[, axis,fill_value])      #右侧小数除法,元素指向
    DataFrame.rtruediv(other[, axis, …])          #右侧真除法,元素指向
    DataFrame.rfloordiv(other[, axis, …])         #右侧向下取整除法,元素指向
    DataFrame.rmod(other[, axis,fill_value])      #右侧模运算,元素指向
    DataFrame.rpow(other[, axis,fill_value])      #右侧幂运算,元素指向
    DataFrame.lt(other[, axis, level])            #类似Array.lt
    DataFrame.gt(other[, axis, level])            #类似Array.gt
    DataFrame.le(other[, axis, level])            #类似Array.le
    DataFrame.ge(other[, axis, level])            #类似Array.ge
    DataFrame.ne(other[, axis, level])            #类似Array.ne
    DataFrame.eq(other[, axis, level])            #类似Array.eq
    DataFrame.combine(other,func[,fill_value, …]) #Add two DataFrame objects and do not propagate NaN values, so if for a
    DataFrame.combine_first(other)                #Combine two DataFrame objects and default to non-null values in frame calling the method.

    函数应用&分组&窗口
    DataFrame.apply(func[, axis, broadcast, …])   #应用函数
    DataFrame.applymap(func)                      #Apply a function to a DataFrame that is intended to operate elementwise, i.e.
    DataFrame.aggregate(func[, axis])             #Aggregate using callable, string, dict, or list of string/callables
    DataFrame.transform(func, *args, **kwargs)    #Call function producing a like-indexed NDFrame
    DataFrame.groupby([by, axis, level, …])       #分组
    DataFrame.rolling(window[, min_periods, …])   #滚动窗口
    DataFrame.expanding([min_periods, freq, …])   #拓展窗口
    DataFrame.ewm([com, span, halflife,  …])      #指数权重窗口

    描述统计学
    DataFrame.abs()                               #返回绝对值
    DataFrame.all([axis, bool_only, skipna])      #Return whether all elements are True over requested axis
    DataFrame.any([axis, bool_only, skipna])      #Return whether any element is True over requested axis
    DataFrame.clip([lower, upper, axis])          #Trim values at input threshold(s).
    DataFrame.clip_lower(threshold[, axis])       #Return copy of the input with values below given value(s) truncated.
    DataFrame.clip_upper(threshold[, axis])       #Return copy of input with values above given value(s) truncated.
    DataFrame.corr([method, min_periods])         #返回本数据框成对列的相关性系数
    DataFrame.corrwith(other[, axis, drop])       #返回不同数据框的相关性
    DataFrame.count([axis, level, numeric_only])  #返回非空元素的个数
    DataFrame.cov([min_periods])                  #计算协方差
    DataFrame.cummax([axis, skipna])              #Return cumulative max over requested axis.
    DataFrame.cummin([axis, skipna])              #Return cumulative minimum over requested axis.
    DataFrame.cumprod([axis, skipna])             #返回累积
    DataFrame.cumsum([axis, skipna])              #返回累和
    DataFrame.describe([percentiles,include, …])  #整体描述数据框
    DataFrame.diff([periods, axis])               #1st discrete difference of object
    DataFrame.eval(expr[, inplace])               #Evaluate an expression in the context of the calling DataFrame instance.
    DataFrame.kurt([axis, skipna, level, …])      #返回无偏峰度Fisher’s  (kurtosis of normal == 0.0).
    DataFrame.mad([axis, skipna, level])          #返回偏差
    DataFrame.max([axis, skipna, level, …])       #返回最大值
    DataFrame.mean([axis, skipna, level, …])      #返回均值
    DataFrame.median([axis, skipna, level, …])    #返回中位数
    DataFrame.min([axis, skipna, level, …])       #返回最小值
    DataFrame.mode([axis, numeric_only])          #返回众数
    DataFrame.pct_change([periods, fill_method])  #返回百分比变化
    DataFrame.prod([axis, skipna, level, …])      #返回连乘积
    DataFrame.quantile([q, axis, numeric_only])   #返回分位数
    DataFrame.rank([axis, method, numeric_only])  #返回数字的排序
    DataFrame.round([decimals])                   #Round a DataFrame to a variable number of decimal places.
    DataFrame.sem([axis, skipna, level, ddof])    #返回无偏标准误
    DataFrame.skew([axis, skipna, level, …])      #返回无偏偏度
    DataFrame.sum([axis, skipna, level, …])       #求和
    DataFrame.std([axis, skipna, level, ddof])    #返回标准误差
    DataFrame.var([axis, skipna, level, ddof])    #返回无偏误差 

    从新索引&选取&标签操作
    DataFrame.add_prefix(prefix)                  #添加前缀
    DataFrame.add_suffix(suffix)                  #添加后缀
    DataFrame.align(other[, join, axis, level])   #Align two object on their axes with the
    DataFrame.drop(labels[, axis, level, …])      #返回删除的列
    DataFrame.drop_duplicates([subset, keep, …])  #Return DataFrame with duplicate rows removed, optionally only
    DataFrame.duplicated([subset, keep])          #Return boolean Series denoting duplicate rows, optionally only
    DataFrame.equals(other)                       #两个数据框是否相同
    DataFrame.filter([items, like, regex, axis])  #过滤特定的子数据框
    DataFrame.first(offset)                       #Convenience method for subsetting initial periods of time series data based on a date offset.
    DataFrame.head([n])                           #返回前n行
    DataFrame.idxmax([axis, skipna])              #Return index of first occurrence of maximum over requested axis.
    DataFrame.idxmin([axis, skipna])              #Return index of first occurrence of minimum over requested axis.
    DataFrame.last(offset)                        #Convenience method for subsetting final periods of time series data based on a date offset.
    DataFrame.reindex([index, columns])           #Conform DataFrame to new index with optional filling logic, placing NA/NaN in locations having no value in the previous index.
    DataFrame.reindex_axis(labels[, axis, …])     #Conform input object to new index with optional filling logic, placing NA/NaN in locations having no value in the previous index.
    DataFrame.reindex_like(other[, method, …])    #Return an object with matching indices to myself.
    DataFrame.rename([index, columns])            #Alter axes input function or functions.
    DataFrame.rename_axis(mapper[, axis, copy])   #Alter index and / or columns using input function or functions.
    DataFrame.reset_index([level, drop, …])       #For DataFrame with multi-level index, return new DataFrame with labeling information in the columns under the index names, defaulting to ‘level_0’, ‘level_1’, etc.
    DataFrame.sample([n, frac, replace, …])       #返回随机抽样
    DataFrame.select(crit[, axis])                #Return data corresponding to axis labels matching criteria
    DataFrame.set_index(keys[, drop, append ])    #Set the DataFrame index (row labels) using one or more existing columns.
    DataFrame.tail([n])                           #返回最后几行
    DataFrame.take(indices[, axis, convert])      #Analogous to ndarray.take
    DataFrame.truncate([before, after, axis ])    #Truncates a sorted NDFrame before and/or after some particular index value.

    处理缺失值
    DataFrame.dropna([axis, how, thresh, …])      #Return object with labels on given axis omitted where alternately any
    DataFrame.fillna([value, method, axis, …])    #填充空值
    DataFrame.replace([to_replace, value, …])     #Replace values given in ‘to_replace’ with ‘value’.

    从新定型&排序&转变形态
    DataFrame.pivot([index, columns, values])     #Reshape data (produce a “pivot” table) based on column values.
    DataFrame.reorder_levels(order[, axis])       #Rearrange index levels using input order.
    DataFrame.sort_values(by[, axis, ascending])  #Sort by the values along either axis
    DataFrame.sort_index([axis, level, …])        #Sort object by labels (along an axis)
    DataFrame.nlargest(n, columns[, keep])        #Get the rows of a DataFrame sorted by the n largest values of columns.
    DataFrame.nsmallest(n, columns[, keep])       #Get the rows of a DataFrame sorted by the n smallest values of columns.
    DataFrame.swaplevel([i, j, axis])             #Swap levels i and j in a MultiIndex on a particular axis
    DataFrame.stack([level, dropna])              #Pivot a level of the (possibly hierarchical) column labels, returning a DataFrame (or Series in the case of an object with a single level of column labels) having a hierarchical index with a new inner-most level of row labels.
    DataFrame.unstack([level, fill_value])        #Pivot a level of the (necessarily hierarchical) index labels, returning a DataFrame having a new level of column labels whose inner-most level consists of the pivoted index labels.
    DataFrame.melt([id_vars, value_vars, …])      #“Unpivots” a DataFrame from wide format to long format, optionally
    DataFrame.T                                   #Transpose index and columns
    DataFrame.to_panel()                          #Transform long (stacked) format (DataFrame) into wide (3D, Panel) format.
    DataFrame.to_xarray()                         #Return an xarray object from the pandas object.
    DataFrame.transpose(*args, **kwargs)          #Transpose index and columns

    Combining& joining&merging
    DataFrame.append(other[, ignore_index, …])    #追加数据
    DataFrame.assign(**kwargs)                    #Assign new columns to a DataFrame, returning a new object (a copy) with all the original columns in addition to the new ones.
    DataFrame.join(other[, on, how, lsuffix, …])  #Join columns with other DataFrame either on index or on a key column.
    DataFrame.merge(right[, how, on, left_on, …]) #Merge DataFrame objects by performing a database-style join operation by columns or indexes.
    DataFrame.update(other[, join, overwrite, …]) #Modify DataFrame in place using non-NA values from passed DataFrame.

    时间序列
    DataFrame.asfreq(freq[, method, how, …])      #将时间序列转换为特定的频次
    DataFrame.asof(where[, subset])               #The last row without any NaN is taken (or the last row without
    DataFrame.shift([periods, freq, axis])        #Shift index by desired number of periods with an optional time freq
    DataFrame.first_valid_index()                 #Return label for first non-NA/null value
    DataFrame.last_valid_index()                  #Return label for last non-NA/null value
    DataFrame.resample(rule[, how, axis, …])      #Convenience method for frequency conversion and resampling of time series.
    DataFrame.to_period([freq, axis, copy])       #Convert DataFrame from DatetimeIndex to PeriodIndex with desired
    DataFrame.to_timestamp([freq, how, axis])     #Cast to DatetimeIndex of timestamps, at beginning of period
    DataFrame.tz_convert(tz[, axis, level, copy]) #Convert tz-aware axis to target time zone.
    DataFrame.tz_localize(tz[, axis, level, …])   #Localize tz-naive TimeSeries to target time zone.

    作图
    DataFrame.plot([x, y, kind, ax, ….])          #DataFrame plotting accessor and method
    DataFrame.plot.area([x, y])                   #面积图Area plot
    DataFrame.plot.bar([x, y])                    #垂直条形图Vertical bar plot
    DataFrame.plot.barh([x, y])                   #水平条形图Horizontal bar plot
    DataFrame.plot.box([by])                      #箱图Boxplot
    DataFrame.plot.density(**kwds)                #核密度Kernel Density Estimate plot
    DataFrame.plot.hexbin(x, y[, C, …])           #Hexbin plot
    DataFrame.plot.hist([by, bins])               #直方图Histogram
    DataFrame.plot.kde(**kwds)                    #核密度Kernel Density Estimate plot
    DataFrame.plot.line([x, y])                   #线图Line plot
    DataFrame.plot.pie([y])                       #饼图Pie chart
    DataFrame.plot.scatter(x, y[, s, c])          #散点图Scatter plot
    DataFrame.boxplot([column, by, ax, …])        #Make a box plot from DataFrame column optionally grouped by some columns or
    DataFrame.hist(data[, column, by, grid, …])   #Draw histogram of the DataFrame’s series using matplotlib / pylab.

    转换为其他格式
    DataFrame.from_csv(path[, header, sep, …])    #Read CSV file (DEPRECATED, please use pandas.read_csv() instead).
    DataFrame.from_dict(data[, orient, dtype])    #Construct DataFrame from dict of array-like or dicts
    DataFrame.from_items(items[,columns,orient])  #Convert (key, value) pairs to DataFrame.
    DataFrame.from_records(data[, index, …])      #Convert structured or record ndarray to DataFrame
    DataFrame.info([verbose, buf, max_cols, …])   #Concise summary of a DataFrame.
    DataFrame.to_pickle(path[, compression, …])   #Pickle (serialize) object to input file path.
    DataFrame.to_csv([path_or_buf, sep, na_rep])  #Write DataFrame to a comma-separated values (csv) file
    DataFrame.to_hdf(path_or_buf, key, **kwargs)  #Write the contained data to an HDF5 file using HDFStore.
    DataFrame.to_sql(name, con[, flavor, …])      #Write records stored in a DataFrame to a SQL database.
    DataFrame.to_dict([orient, into])             #Convert DataFrame to dictionary.
    DataFrame.to_excel(excel_writer[, …])         #Write DataFrame to an excel sheet
    DataFrame.to_json([path_or_buf, orient, …])   #Convert the object to a JSON string.
    DataFrame.to_html([buf, columns, col_space])  #Render a DataFrame as an HTML table.
    DataFrame.to_feather(fname)                   #write out the binary feather-format for DataFrames
    DataFrame.to_latex([buf, columns, …])         #Render an object to a tabular environment table.
    DataFrame.to_stata(fname[, convert_dates, …]) #A class for writing Stata binary dta files from array-like objects
    DataFrame.to_msgpack([path_or_buf, encoding]) #msgpack (serialize) object to input file path
    DataFrame.to_sparse([fill_value, kind])       #Convert to SparseDataFrame
    DataFrame.to_dense()                          #Return dense representation of NDFrame (as opposed to sparse)
    DataFrame.to_string([buf, columns, …])        #Render a DataFrame to a console-friendly tabular output.
    DataFrame.to_clipboard([excel, sep])          #Attempt to write text representation of object to the system clipboard 
    --------------------- 
    作者:brucewong0516 
    来源:CSDN 
    原文:https://blog.csdn.net/brucewong0516/article/details/81782312 

  • 相关阅读:
    python函数完整语法和分类
    python数据类型int、float、str、list、dict、set定义及常用方法汇总
    python语言之流程控制
    python用户交互、基本数据类型、运算符
    BZOJ 3925: [Zjoi2015] 地震后的幻想乡(概率DP)
    BZOJ 3566 概率充电器(树形概率DP)
    BZOJ 1778: [Usaco2010 Hol]Dotp 驱逐猪猡 (高斯消元)
    BZOJ 3589 动态树 (树链剖分+线段树)
    BZOJ 3672[NOI2014]购票(树链剖分+线段树维护凸包+斜率优化) + BZOJ 2402 陶陶的难题II (树链剖分+线段树维护凸包+分数规划+斜率优化)
    BZOJ 4013/Luogu P3240 [HNOI2015] 实验比较 (树形DP)
  • 原文地址:https://www.cnblogs.com/cxxBoo/p/13852940.html
Copyright © 2011-2022 走看看