--- title: numpy，pandas，matplotlib库的使用 date: 2025-03-07 19:10:52 categories: - 人工智能基础 --- # 前言： Python三个常用库的学习笔记 # 一、numpy ## 1.1 数组创建 **创建函数：array()** 几个参数： object：默认参数，输入一个列表，列表中又可以嵌套列表来表示多维数据 dtype:指定数据类型，可选，如果不指定则自动识别 order：指定存储方式，C-按行（先存储第一行的所有元素，然后是第二行的所有元素） F-按列（先存储第一列的所有元素，然后是第二列的所有元素，依此类推） ndmin: 设置数组维度，一般嵌套有几层列表就是几维 ```python a = np.array([[1,2,3],[4,5,6]], dtype = np.float32, order = "F", ndmin = 3) ``` **运行结果：** array([[[1., 2., 3.], [4., 5., 6.]]], dtype=float32) **打印数组形状** 返回一个元组（行，列） ```python a = np.array([[1,2,3],[4,5,6],[7,8,9]]) print(a.shape) ``` **运行结果：** (3, 3) **填充数组：** zeros(（行，列）):根据指定形状创建数组，元素默认以0填充 ```python a = np.zeros((3,3)) a ``` **运行结果：** array([[0., 0., 0.], [0., 0., 0.], [0., 0., 0.]]) ones(（行，列）):根据指定的形状创建数组，元素默认以1填充 ```python a = np.ones((3,4)) a ``` **运行结果：** array([[1., 1., 1., 1.], [1., 1., 1., 1.], [1., 1., 1., 1.]]) full(（行，列），fill_value = 填充值):根据指定的形状创建数组，元素可以使用指定值填充 ```python a = np.full((3,3),fill_value=5) a ``` **运行结果：** array([[5, 5, 5], [5, 5, 5], [5, 5, 5]]) arange(start,stop,step):根据参数创建一个一维的等差数组 ```python a = np.arange(0,10,2) print(a) ``` linspace():生成一个等差数列的一维数组 start:起始值 stop:终止值 num:要分成的份数，默认是50 endpoint:默认为True，表示包含stop 每一份之间的间隔值：（stop - start）/num - 1 ```python a = np.linspace(0,10,21) print(a) ``` **运行结果：** [ 0. 0.5 1. 1.5 2. 2.5 3. 3.5 4. 4.5 5. 5.5 6. 6.5 7. 7.5 8. 8.5 9. 9.5 10. ] ## 1.2 数组属性 **shape:** 1.返回一个形状元祖 2.直接修改数组形状，数组的元素不能改变，既行×列＝元素个数 ```python a = np.array([1,2,3,4,5,6,7,8,9,10,11,12]) print(a) print(a.shape) a.shape = (2,6) print(a) print(a.shape) a.shape = (3,4) print(a) print(a.shape) ``` **运行结果：** [ 1 2 3 4 5 6 7 8 9 10 11 12] (12,) [[ 1 2 3 4 5 6] [ 7 8 9 10 11 12]] (2, 6) [[ 1 2 3 4] [ 5 6 7 8] [ 9 10 11 12]] (3, 4) **ndim:** 返回数组维度数 ```python a = np.array([1,2,3,4,5,6,7,8,9,10,11,12]) print(a.ndim) ``` **运行结果：** 1 ## 1.3 数组索引 **切片索引：** (行start:行stop:行step,列start:列stop:列step)stop是不包含的 numpy数组切片可以按行也可以按列 ```python a = np.array([[1,2,3],[4,5,6],[7,8,9]]) print(a[:,0]) # ...表示保留整行/列数据,相当于：： print(a[...,0]) # 查询第一行第一二列数据 print(a[0,0:2]) # 查询第二三行，第二三列数据 print(a[1:3,1:3]) # 查询第二行，第二列的数据 print(a[1,1]) # 查询第二三行数 print(a[1:3,...]) ``` **运行结果:** [1 4 7] [1 4 7] [1 2] [[5 6] [8 9]] 5 [[4 5 6] [7 8 9]] **整数数组索引:** 数组正常创建，在切片的时候a[[行索引],[列索引]]，行索引和列索引组合成为坐标定位到数组元素 ```python a = np.array([[1, 2, 3], [4, 5, 6]]) print("整数数组索引的结果：", a[[0, 0, 1], [1, 2, 0]]) ``` **运行结果：** 整数数组索引的结果： [2 3 4] **布尔索引 ：** 使用条件表达式来访问满足条件的元素，二维数组布尔索引直接返回满足条件的一维数组 ```python a = np.array([[[1,2,3],[4,5,6]]]) bool_index = a>2 print(bool_index) result = a[bool_index] print(result) ``` **运行结果：** [[[False False True] [ True True True]]] [3 4 5 6] 也可以再细化判断条件，例如索引第3个元素大于3的所有行 ```python print(arr[arr[:, 2] > 6]) ``` 索引第2行 值大于3 的所有的元素 所在的列 ```python print(arr[:, arr[1] > 5]) ``` ## 1.4 数组广播 目的：把两个不同形状的数组变成相同形状 手段：将数组的形状较小的数组复制到数组较大的数组的对应位置，使得它们具有相同的形状。 广播的前提：两个数组的维度相同，或者其中一个数组的维度为1。 维度：维度是指数组中元素的数量，数组的维度越高，数组中元素的数量就越多。 ```python import numpy as np a = np.array([[1], [2],[3]]) print(a.shape) # 3×1的数组加上1×3的数组 b = np.array([10,20,30]) print(b.shape) c = a + b print(c) a = np.array([1,2,3]) b = np.array([[10],[20],[30]]) c = a + b print(c) ``` **运行结果：** (3, 1) (3,) [[11 21 31] [12 22 32] [13 23 33]] [[11 12 13] [21 22 23] [31 32 33]] ## 1.5 数组遍历 for循环遍历： 二维数组直接使用for循环遍历，输出结果是每一个一维数组，有几维就要写几层for循环，效率不高 ```python a = np.array([[1, 2, 3], [4, 5, 6]]) for row in a: print(row) ``` **运行结果：** [1 2 3] [4 5 6] **nditer()函数** 可以对数组进行迭代，并返回一个迭代器对象。 控制参数： 1.order:设置遍历数组的顺序，C-按行优先，F-按列优先。 ```python a = np.array([[1,2,3],[4,5,6]]) for i in np.nditer(a,order="C"): print(i,end = " ") print("=====") for i in np.nditer(a,order="F"): print(i,end = " ") ``` 1 2 3 4 5 6 ===== 1 4 2 5 3 6 flags（[参数]）：指定迭代器的额外行为 参数值： multi_index:返回元素对应的下标索引元组(行，列) external_loop:将遍历的单个元素添加到一个一维数组，遍历完成后输出一维数组 ```python a = np.array([[1,2,3],[4,5,6]]) it = np.nditer(a,flags = ["multi_index"]) for i in it: print(it.multi_index) b = np.array([[7,8,9],[10,11,12]]) for x in np.nditer(b, flags=['external_loop'], order='C'): print(x) ``` [1 2 3 4 5 6] ## 1.6 数组运算 **dot** **一维数组点积** ```python a = np.array([1, 2, 3]) b = np.array([4, 5, 6]) print(np.dot(a, b)) ``` 1×4＋2×5＋3×6 = 32 **二维数组矩阵乘法** ```python a = np.array([[1, 2], [3, 4]]) b = np.array([[5, 6], [7, 8]]) c = np.dot(a, b) print(c) ``` [[19 22] [43 50]] **matmul()** **两个矩阵的乘法运算** ```python a = np.array([[1, 2], [3, 4],[9,10]]) b = np.array([[5, 6], [7, 8]]) c = np.matmul(a, b) print(c) ``` [[ 19 22] [ 43 50] [115 134]] **np.linalg.det（）** 计算一个方阵的行列式 ```python a = np.array([[1, 2, 3], [4, 8, 6], [7, 8, 9]]) print(np.linalg.det(a)) ``` -36.0 ## 1.7 数组操作 reshape((行，列))：修改数组形状 修改后返回一个新数组，不直接返回原数组 返回的新数组是原数组的一个视图，修改视图会影响原数组 reshape和shape的区别：reshape是改变数组的形状，shape是查看数组的形状 ```python a = np.array([1,2,3,4,5,6]) a1 = a.reshape((2,3)) print(a1) # 修改视图 a1[0][0] = 100 print(a) print(a1) ``` [[1 2 3] [4 5 6]] [100 2 3 4 5 6] [[100 2 3] [ 4 5 6]] 参数： -1占位符：numpy会自动计算该占位符的维度 ```python a = np.array([1,2,3,4,5,6]) print(a) a1 = a.reshape((3,-1)) print(a1) # 等价于 a1 = a.reshape((3,2)) print(a1) ``` [1 2 3 4 5 6] [[1 2] [3 4] [5 6]] [[1 2] [3 4] [5 6]] np.resize(arr,(x,y)):重塑数组形状，但是没有元素个数限制,直遍历原数组填充新数组直至填满新数组,剩下未遍历完的数组丢弃 ```python a = np.arange(1,13).reshape(3,4) a1 = np.resize(a,(6,5)) print(a1) ``` [[ 1 2 3 4 5] [ 6 7 8 9 10] [11 12 1 2 3] [ 4 5 6 7 8] [ 9 10 11 12 1] [ 2 3 4 5 6]] np.expand_dims(arr,axis):根据指定的轴方向进行升维 axis = 0,按行升维，axis = 1,按列升维 ```python a = np.array([1,2,3]) a1 = np.expand_dims(a,axis = 0) print(a1) print(np.shape(a1)) a2 = np.expand_dims(a,axis = 1) print(a2) print(np.shape(a2)) ``` [[1 2 3]] (1, 3) [[1] [2] [3]] (3, 1) **多维数组扁平化** **flat属性**：返回一个一维数组迭代器，可以使用循环来遍历 ```python a = np.arange(1,13).reshape(3,4) for i in a.flat: print(i,end=' ') print() ``` 1 2 3 4 5 6 7 8 9 10 11 12 **flatten()方法**：返回一个一维数组 浅拷贝，是原数组的副本，修改新数组不会影响原数组 ```python a = np.arange(1,13).reshape(3,4) b = a.flatten() print(b) b[0] = 100 print(a) ``` [ 1 2 3 4 5 6 7 8 9 10 11 12] [[ 1 2 3 4] [ 5 6 7 8] [ 9 10 11 12]] **ravel()方法**：返回一个一维数组 深拷贝，是原数组的试图，修改新数组会影响原数组 ```python a = np.arange(1,13).reshape(3,4) b = a.ravel() print(b) b[0] = 100 print(a) ``` [ 1 2 3 4 5 6 7 8 9 10 11 12] [[100 2 3 4] [ 5 6 7 8] [ 9 10 11 12]] **数组转置** ```python a = np.array([[1, 2, 3], [4, 5, 6]]) print(a.T) ``` [[1 4] [2 5] [3 6]] **squeeze(arr,axis):**根据指定的轴进行降维 降维前提：所在轴上的维度必须为1才能进行降维 axis按照轴方向进行降维，0-按最外层降维，1-按第二层降维，以此类推 假设数组为二维数组 axis = 0，按行降维 axis = 1，按列降维 axis = None，或不指定axis，对所有维度数为1的项降维 移除元素个数为1的维度 ```python a = np.array([[[[[1,2,3],[4,5,6],[7,8,9]],[[10,11,12],[13,14,15],[16,17,18]]]]]) print(a.shape) # 按照第一层降维 a1 = np.squeeze(a,axis=0) print(a1) print(a1.shape) # 按照第二层降维 a2 = np.squeeze(a,axis=1) print(a2) print(a2.shape) ``` (1, 1, 2, 3, 3) [[[[ 1 2 3] [ 4 5 6] [ 7 8 9]] [[10 11 12] [13 14 15] [16 17 18]]]] (1, 2, 3, 3) [[[[ 1 2 3] [ 4 5 6] [ 7 8 9]] [[10 11 12] [13 14 15] [16 17 18]]]] (1, 2, 3, 3) **数组拼接stack** hstack(tuple):接收一个包含要连接元素的元组，**横向连接**，要求行数相同 ```python a = np.array([[1,2],[3,4]]) b = np.array([[5],[6]]) c = np.hstack((a,b)) print(c) ``` [[1 2 5] [3 4 6]] vstack(tuple):接收一个包含要连接元素的元组，**纵向连接**，要求列数相同 ```python a = np.array([[1,2,3],[4,5,6]]) b = np.array([7,8,9]) c = np.array([[10,11,12],[13,14,15]]) d = np.vstack((a,b,c)) print(d) ``` [[ 1 2 3] [ 4 5 6] [ 7 8 9] [10 11 12] [13 14 15]] **数组切割split** hsplit(arr,[index]):**水平切割数组** ```python a = np.arange(1,13).reshape(3,4) print(a) a1 = np.hsplit(a,[1,3]) for ai in a1: print("===========") print(ai) print("===========") ``` [[ 1 2 3 4] [ 5 6 7 8] [ 9 10 11 12]] [[1] [5] [9]] =========== [[ 2 3] [ 6 7] [10 11]] =========== [[ 4] [ 8] [12]] vsplit(arr,[index]):**竖直切割数组** ```python a2 = np.vsplit(a,[1,2]) for ai in a2: print("===========") print(ai) print("===========") ``` =========== [[1 2 3 4]] =========== [[5 6 7 8]] =========== [[ 9 10 11 12]] **append()**:在数组**尾部**添加值（可以是数组） axis = None，添加后**整合成一维数组输出** ```python a = np.array([[1,2,3],[4,5,6]]) a1 = np.append(a,[1,1,1],axis = None) print(a1) ``` axis = 0**添加到行**，维度必须保持一致 ```python a2 = np.append(a,[[1,1,8]],axis = 0) print(a2) ``` axis = 1**添加到列**，维度必须保持一致 ```python a3 = np.append(a,[[1,1,1],[4,5,6]],axis = 1) print(a3) ``` [1 2 3 4 5 6 1 1 1] [[1 2 3] [4 5 6] [1 1 8]] [[1 2 3 1 1 1] [4 5 6 4 5 6]] **insert()**指定位置添加值 axis = None,返回一维数组 axis = 0,插入行，自动广播 axis = 1,插入列，自动广播 ```python a = np.array([[1,2,3],[4,5,6]]) a1 = np.insert(a,1,[6],axis = None) print(a1) a2 = np.insert(a,1,[6],axis = 0) print(a2) a3 = np.insert(a,1,[6],axis = 1) print(a3) ``` [1 6 2 3 4 5 6] [[1 2 3] [6 6 6] [4 5 6]] [[1 6 2 3] [4 6 5 6]] **delete()：**删除指定位置的元素 参数和append类似 ```python a = np.arange(1,13).reshape(3,4) a1 = np.delete(a,1,axis = None) print(a1) a = np.arange(1,13).reshape(3,4) a2 = np.delete(a,1,axis = 0) print(a2) a = np.arange(1,13).reshape(3,4) a3 = np.delete(a,1,axis = 1) print(a3) ``` [ 1 3 4 5 6 7 8 9 10 11 12] [[ 1 2 3 4] [ 9 10 11 12]] [[ 1 3 4] [ 5 7 8] [ 9 11 12]] **argwhere()**:默认返回非0元素的索引坐标，可自己设置条件 ```python a = np.array([[0,1,0,2],[0,2,0,3]]) a1 = np.argwhere(a) print(a1) ``` [[0 1] [0 3] [1 1] [1 3]] **where**和argwhere功能一样，但是返回的是元组 (行索引下标数组，列索引下标数组) 可以整合布尔数组和索引数组，实现复杂的条件筛选 ```python a = np.array([[0,1,0,2],[0,2,0,3]]) a1 = np.where(a) print(a1) a2 = np.where(a>1) print(a2) a3 = np.where(a>0,a,9)#将小于等于0的元素替换为9 print(a3) ``` (array([0, 0, 1, 1], dtype=int64), array([1, 3, 1, 3], dtype=int64)) (array([0, 1, 1], dtype=int64), array([3, 1, 3], dtype=int64)) [[9 1 9 2] [9 2 9 3]] **argmax()**获取数组中第一个最大值对应的**下标索引** ```python a = np.array([[0,3,0,2],[0,2,0,3]]) a1 = np.argmax(a,axis=0) print(a1) ``` [0 0 0 1] **unique()函数：**返回数组中唯一的元素 参数： arr：输入数组 return_index：如果为True，返回**索引数组**；否则，返回值**数组**。 return_inverse：如果为True，返回**反向映射**(反向映射：元素到其在数组中出现的索引)数组；否则，返回**None**。 return_counts：如果为True，返回**每个元素的计数**；否则，返回**None**。 ```python arr = np.array([1, 2, 3, 2, 4, 1, 5]) print(np.unique(arr)) # [1 2 3 4 5] a,index = np.unique(arr,return_index=True) print("数组：",a,"\n索引：",index) a,index,inverse = np.unique(arr,return_index=True,return_inverse=True) print("数组：",a,"反向映射：",inverse) a,index,inverse,count = np.unique(arr,return_index=True,return_inverse=True,return_counts=True) print("数组：",a,"计数：",count) ``` [1 2 3 4 5] 数组： [1 2 3 4 5] 索引： [0 1 2 4 6] 数组： [1 2 3 4 5] 反向映射： [0 1 2 1 3 0 4] 数组： [1 2 3 4 5] 计数： [2 2 1 1 1] **amin()和amax()函数** 对于二维数组来说， axis = 0表示沿着竖直的方向求最小值或最大值， axis = 1表示沿着水平的方向求最小值或最大值。 axis = None表示沿着整个数组的方向求最小值或最大值。 ```python a = np.array([[0, 7, 3], [7, 9, 6], [4, 2, 9]]) print(np.amin(a,axis=0)) # 每一列的最小值 print(np.amax(a,axis=1)) # 每一行的最大值 print(np.amin(a,axis=None)) # 整个数组的最小值 ``` [0 2 3] [7 9 9] 0 **ptp()计算峰值的差** ```python arr = np.array([[1,6,8],[0,9,4],[5,6,9]]) a1 = np.ptp(arr)#整个数组的最大值和最小值相减 print(a1) a2 = np.ptp(arr,axis = 0)#数组每一列的最大值和最小值相减 print(a2) a3 = np.ptp(arr,axis = 1)#数组每一行的最大值和最小值相减 print(a3) ``` 9 [5 3 5] [7 9 4] **median()计算中位数** 参数axis和ptp一样，都可以单独按行或者按列操作 中位数计算方法和数学一样（特殊情况取平均） ```python # 中位数计算方法和数学一样（特殊情况取平均） arr = np.array([[1,2,0],[0,9,4],[5,6,9]]) a1 = np.median(arr) print(a1) a2 = np.median(arr,axis = 0) print(a2) a3 = np.median(arr,axis = 1) print(a3) ``` 4.0 [1. 6. 4.] [1. 4. 6.] **mean()计算算术平均值** ```python arr = np.array([[1,2,0],[0,9,4],[5,6,9]]) a1 = np.mean(arr) print(a1) a2 = np.mean(arr,axis=0) print(a2) a3 = np.mean(arr,axis = 1) print(a3) ``` 4.0 [2. 5.66666667 4.33333333] [1. 4.33333333 6.66666667] **average()加权平均值** ```python arr = np.array([1,2,6]) weight = np.array([0.1,0.2,0.7]) a1 = np.average(arr,weights=weight) # 必须使用关键字参数 weights print(a1) ``` 4.699999999999999 **二维数组的加权平均值** axis=None表示沿着所有维度求平均值 axis=0表示沿着行方向求平均值，axis=1表示沿着列方向求平均值 weights的维度必须与数组arr的维度相同，且每一行的权重之和必须为1 ```python arr2 = np.array([[1,2,3],[4,5,6]]) weight2 = np.array([[0.1,0.2,0.2],[0.1,0.1,0.3]]) a2 = np.average(arr2,axis=None,weights=weight2) print(a2) a3 = np.average(arr2,axis=0,weights=weight2) print(a3) a4 = np.average(arr2,axis=1,weights=weight2) print(a4) ``` 3.8 [2.5 3. 4.8] [2.2 5.4] **var()方差** ```python a1 = np.array([1, 2, 3, 4, 5,6]) a2 = np.array([[1,2,3],[4,5,6]]) print(np.var(a1)) print(np.var(a2,axis=0)) print(np.var(a2,axis=1)) ``` 2.9166666666666665 [2.25 2.25 2.25] [0.66666667 0.66666667] 在样本数据中，样本均值的估计会引入一定的偏差。通过使用 n−1作为分母，可以校正这种偏差，得到更准确的总体方差估计。 ```python arr = np.array([1, 2, 3, 4, 5]) print(np.var(arr, ddof=1)) ``` 2.5 **std()标准差** ```python a = np.array([1, 2, 3, 4, 5, 6]) print(np.std(a)) a1 = np.array([[1, 2, 3], [4, 5, 6]]) print(np.std(a1, axis=0)) print(np.std(a1, axis=1)) ``` 1.707825127659933 [1.5 1.5 1.5] [0.81649658 0.81649658] # 二、matplotlib ## 2.1 画图操作 **第一种绘图方式，直接使用plot()绘图** plot 绘制曲线函数 这是最简单的绘制方式，常用于简单的查看数据分布 ```python def test01(): x = np.linspace(-6,6,100) y = np.sin(x) z = np.cos(x) plt.plot(x,y,"b") plt.plot(x,z,"r") plt.show() test01() ``` ![img](2025-08-07-numpy，pandas，matplotlib库的使用/557b69e510e246ab8917e7ca864995fd.png) 下面是创建画布和子图的操作： **subplots()函数创建多个子图** ```python def test03(): x = np.linspace(-2,2,100) y = x ** 2 z = x ** -1 m = np.sin(x) n = np.cos(x) # 为避免频繁调用add_subplot，subplot(nrows, ncols) # nrows: 行数 # ncols: 列数 # 返回的对象：fig 画布对象，ax 子图对象列表,该对象是一个数组 fig,ax = plt.subplots(1,4) # 在绘图区域上分别绘图 ax[0].plot(x,y,label = "y = x^2") ax[1].plot(x,z,label = "z = x^-1") ax[2].plot(x,m,label = "m = sin(x)") ax[3].plot(x,n,label = "n = cos(x)") fig,ax = plt.subplots(2,2) #可以随意改变形状，但是子图区域必须一致 ax[0][0].plot(x,y,"r",label = "y = x^2") ax[0][1].plot(x,z,"b",label = "z = x^-1") ax[1][0].plot(x,m,"g",label = "m = sin(x)") ax[1][1].plot(x,n,"y",label = "n = cos(x)") plt.show() test03() ``` ![img](2025-08-07-numpy，pandas，matplotlib库的使用/6d920da7363f4946abfd04bff5876925.png) ![img](2025-08-07-numpy，pandas，matplotlib库的使用/898183689100490087eb592f41094ae3.png) ## 2.2 绘制图表 **柱状图bar()** ```python x = ["A","B","C","D"] y = [20,30,40,50] y2 = [30,40,50,60] # 参数： # height：y轴数据 # bottom：X轴数据 # align:刻度在柱子的位置：center，left，right fig,ax = plt.subplots() # 绘制第二个数据集的柱状图，堆叠在第一个数据集上 ax.bar(x, y2, color='lightgreen', align="center") ax.bar(x,y,color = "skyblue",align="center") ax.set_title('Customized Bar Chart') ax.set_xlabel('Categories') ax.set_ylabel('Values') plt.show() ``` ![img](2025-08-07-numpy，pandas，matplotlib库的使用/eb46cb1bbe30456293985903781fb5f6.png) 直方图hist ```python data = np.random.randn(1000) fix,ax = plt.subplots() ax.hist(data, bins=30, color='skyblue', edgecolor='black') plt.show() ``` ![img](2025-08-07-numpy，pandas，matplotlib库的使用/8b0f43aa35e24f06bdee3bb49742ffa7.png) **饼图ple()** ```python labels = ['A', 'B', 'C', 'D'] sizes = [15, 30, 45, 10] fig, ax = plt.subplots() ax.pie(sizes, labels=labels, autopct='%1.1f%%', startangle=90) # 设置标题 ax.set_title('Simple Pie Chart') # 显示图形 plt.show() ``` ![img](2025-08-07-numpy，pandas，matplotlib库的使用/947c86270b244d5099adfa9018d6930f.png) **折线图plot** ```python x = np.linspace(0, 10, 100) y1 = np.sin(x) y2 = np.cos(x) # 创建图形和子图 fig, ax = plt.subplots() # 绘制多条折线图 ax.plot(x, y1, label='sin(x)', color='blue') ax.plot(x, y2, label='cos(x)', color='red') # 设置标题和标签 ax.set_title('Multiple Line Charts') ax.set_xlabel('X-axis') ax.set_ylabel('Y-axis') # 添加图例 ax.legend() # 显示图形 plt.show() ``` ![img](2025-08-07-numpy，pandas，matplotlib库的使用/2f63f3962b0e4ea887a296dbbd5b42e2.png) **散点图scatter（）** ```python plt.rcParams['font.sans-serif'] = ['SimHei'] plt.rcParams['axes.unicode_minus'] = False fig = plt.figure() axes = fig.add_axes([.1,.1,.8,.8]) x = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun'] data = [ [120, 132, 101, 134, 90, 230, 210], [220, 182, 191, 234, 290, 330, 310], ] y0 = data[0] y1 = data[1] axes.scatter(x,y0,color='red') axes.scatter(x,y1,color='blue') axes.set_title('散点图') axes.set_xlabel('日期') axes.set_ylabel('数量') plt.legend(labels=['Email', 'Union Ads'],) plt.show() ``` ![img](2025-08-07-numpy，pandas，matplotlib库的使用/9cadcd24dff44d0d9c2b41368a56e2b6.png) # 三、pandas库 ## 3.1 series创建 创建Series对象 ```python s = pd.Series([1, 2, 3, 4, 5],dtype = "f8") print(s) ``` 指定索引和名称： ```python # 创建Series对象，并指定索引 s = pd.Series([1, 2, 3, 4, 5], index=['a', 'b', 'c', 'd', 'e']) print(s.index) # 创建Series对象，并指定索引和名称 s = pd.Series([1, 2, 3, 4, 5], index=['a', 'b', 'c', 'd', 'e'], name='my_series') print(s.name) ``` 使用数组创建Series ```python array_one = np.array(["xiaoming","xiaohuang","xiaozhang"]) series_one = pd.Series(data = array_one,index = [i+1 for i in range(len(array_one))]) print(series_one) ``` ## 3.2 Series遍历 切片 ```python s = pd.Series([1, 2, 3, 4, 5],index = ['a', 'b', 'c', 'd', 'e']) print(s["a"]) # 使用下标切片，终止值不被包含 print(s[0:3]) # 使用标签切片，终止值被包含 print(s['a':'d']) ``` 使用Series的index属性： ```python s = pd.Series([1, 2, 3, 4, 5],index = ['a', 'b', 'c', 'd', 'e']) for index in s.index: print("index:",index,"value:",s[index]) ``` values属性： ```python s = pd.Series([1, 2, 3, 4, 5],index = ['a', 'b', 'c', 'd', 'e']) for value in s.values: print("values:",value) ``` ## 3.3 dataframe创建 **创建dataframe空对象** ```python df = pd.DataFrame() print(df) ``` **通过嵌套字典创建** ```python l = [{"name":"zhangsan","age":20},{"name":"lisi","age":"30","sex":"boy"}] df = pd.DataFrame(l) print(df) ``` **字典嵌套Series创建,如果第二列比第一列多一个元素，自动用nan填充第一列** ```python dic = {"name":pd.Series([1,2,3],index = ["a","b","c"]), "age":pd.Series([10,20,30,40],index = ["a","b","c","d"])} df = pd.DataFrame(dic) print(dic) ``` ## 3.4 dataframe列操作 ```python dic = {"one":[1,2,3],"two":[4,5,6],"three":[7,8,9]} df = pd.DataFrame(dic) print(df) # 取一列返回Series print(df["one"]) # 取多列返回dataframe print(df[["one","two"]]) ``` one two three 0 1 4 7 1 2 5 8 2 3 6 9 0 1 1 2 2 3 Name: one, dtype: int64 one two 0 1 4 1 2 5 2 3 6 **添加空列：** ```python data = {"one":pd.Series(data = [1,2,3],index = ["a","b","c"]), "two":pd.Series(data = [1,2,3,4],index = ["a","b","c","d"])} df = pd.DataFrame(data) print(df) ``` **添加一列空值：** ```python df["Three"] = None print(df) ``` **通过assign()添加一列,等号左边是要添加的列名，右边是对应的列值** **assign()是链式调用** ```python df1 = df.assign(four = [40,50,60,70]).assign(five = [1,2,3,4]) print(df1) ``` one two a 1.0 1 b 2.0 2 c 3.0 3 d NaN 4 one two Three a 1.0 1 None b 2.0 2 None c 3.0 3 None d NaN 4 None one two Three four five a 1.0 1 None 40 1 b 2.0 2 None 50 2 c 3.0 3 None 60 3 d NaN 4 None 70 4 insert()在指定位置插入数据 **参数：loc-要插入的索引下标，column-要插入的列名，value-要插入的数据** ```python data = { "one":pd.Series(data = [1,2,3],index = ["a","b","c"]), "two":pd.Series(data = [7,4,1],index = ["a","b","c"]), "three":pd.Series(data = [2,4,6],index = ["a","b","c"]) } df = pd.DataFrame(data) df.insert(loc=1,column="four",value=[5,6,7]) print(df) ``` **修改数据** ```python data = { "one":pd.Series(data = [1,2,3],index = ["a","b","c"]), "two":pd.Series(data = [7,4,1],index = ["a","b","c"]), "three":pd.Series(data = [2,4,6],index = ["a","b","c"]) } df = pd.DataFrame(data) print(df) print("======================") ``` **修改列名** ```python df.columns = ["col1","col2","col3"] print(df) print("======================") ``` **修改列值(这里用了广播机制)** ```python df["col1"] += 100 print(df) print("======================") ``` **rename()返回一个新的DataFrame对象，不修改原对象** **columns参数可以传入一个字典，将原列名映射到新列名** ```python de1 = df.rename(columns={"col1":"col4","col2":"col5","col3":"col6"}) print(de1) ``` one two three a 1 7 2 b 2 4 4 c 3 1 6 col1 col2 col3 a 1 7 2 b 2 4 4 c 3 1 6 col1 col2 col3 a 101 7 2 b 102 4 4 c 103 1 6 col4 col5 col6 a 101 7 2 b 102 4 4 c 103 1 6 **修改数据类型** ```python data = { "one":pd.Series(data = [1,2,3],index = ["a","b","c"]), "two":pd.Series(data = [7,4,1],index = ["a","b","c"]), "three":pd.Series(data = [2,4,6],index = ["a","b","c"]) } df = pd.DataFrame(data) print(df.dtypes) print("="*50) df["one"] = df["one"].astype(float) df["two"] = df["two"].astype(int) df["three"] = df["three"].astype(str) print(df.dtypes) ``` **删除数据drop()** 参数 labels:要删除的标签，可以是行，也可以是列 axis：轴方向，和labels结合使用 index：要删除的行标签或列标签列表 columns:要删除的列标签或列表 inplaces:如果为True表示原地删除数据，False则是返回一个新的dataframe **删除列 "two"，返回一个新的 DataFrame** ```python df1 = df.drop(labels="two", axis=1, inplace=False) print("df1:") print(df1) ``` **删除索引 "a" 和 "b"，并删除列 "one"，直接在原 DataFrame 上修改** ```python df.drop(index=["a", "b"], columns=["one"], inplace=True) print("df:") print(df) ``` ## 3.5 行操作 **获取a行数据** ```python data = { "one": pd.Series(data=[1, 2, 3], index=["a", "b", "c"]), "two": pd.Series(data=[7, 4, 1], index=["a", "b", "c"]), "three": pd.Series(data=[2, 4, 6], index=["a", "b", "c"]) } df = pd.DataFrame(data) print(df.loc["a"]) ``` **对行切片，获取a行到b行的数据** ```python print(df.loc["a":"b"]) ``` **对行和列切片** ```python print(df.loc["a":"b","one":"two"]) ``` **定位获取一个标量** ```python print(df.loc["a","one"]) ``` **选择获取多行多列** ```python print(df.loc[["a","c"],["one","three"]]) ``` **iloc：根据索引选择数据** ```python data = { 'A': [1, 2, 3, 4], 'B': [5, 6, 7, 8], 'C': [9, 10, 11, 12] } df = pd.DataFrame(data, index=['a', 'b', 'c', 'd']) df.columns = ["1", "2", "3"] df.index = ["A","B","C","D"] print(df) # 使用 iloc 选择数据，先行后列 print(df.iloc[0]) # 选择第 0 行的数据 print(df.iloc[0:2]) # 选择第 0 到 1 行的数据 print(df.iloc[0, 1]) # 选择第 0 行，第 1 列的元素 print(df.iloc[[0, 2], [0, 2]]) # 选择第 0 和 2 行，第 0 和 2 列的数据 print(df.iloc[:,[0,2]]) # 选择第 0 和 2 列的数据 ``` **添加新行,只能有loc，不能用iloc** ```python data = { 'A': [1, 2, 3, 4], 'B': [5, 6, 7, 8], 'C': [9, 10, 11, 12] } df = pd.DataFrame(data,index=["a","b","c","d"]) df.loc["e"] = [17,18,19] df.loc[:,"D"] = [13,14,15,16,17] df ``` **concat拼接dataframe** ```python data1 = { 'A': [1, 2, 3, 4], 'B': [5, 6, 7, 8], 'C': [9, 10, 11, 12] } data2 = { 'B': [5,6,7,8], 'C': [7,8,9,10], 'G': [0,1,0,1,] } df1 = pd.DataFrame(data1,index = ["a","b","c","d"]) df2 = pd.DataFrame(data2,index = ["a","b","c","e"]) print(df1) print(df2) # ignore_index如果为True，则忽略原来的标签重新生成,否则标签中可能有重复名字 df3 = pd.concat([df1,df2],axis=1,ignore_index=False) print(df3) print("_"*50) # inner表示交集 df4 = pd.concat([df1,df2],axis=0,ignore_index=False,join="inner") print(df4) print("_"*50) # 默认outer拼接，并集 df5 = pd.concat([df1,df2],axis=0,ignore_index=False,join="outer") print(df5) ``` **3.6 函数操作** ```python data = { 'A': [1, 2, 3, 4, 5], 'B': [10, 20, 30, 40, 50], 'C': [100, 200, 300, 400, 500] } df = pd.DataFrame(data) # 计算每列的均值 mean_values = df.mean() print(mean_values) # 计算每列的中位数 median_values = df.median() print(median_values) #计算每列的方差 var_values = df.var() print(var_values) # 计算每列的标准差 std_values = df.std() print(std_values) # 计算每列的最小值 min_values = df.min() print("最小值:") print(min_values) # 计算每列的最大值 max_values = df.max() print("最大值:") print(max_values) # 计算每列的总和 sum_values = df.sum() print(sum_values) # 计算每列的非空值数量 count_values = df.count() print(count_values) ``` **reindex重置索引** ```python data = { 'A': [1, 2, 3, 4, 5], 'B': [10, 20, 30, 40, 50], 'C': [100, 200, 300, 400, 500] } df = pd.DataFrame(data) print("原始 DataFrame:") print(df) # 使用 reindex 方法重置行索引 new_index = ["b", "c", "d", "e", "f"] df_reindexed = df.reindex(index = new_index) print("\n使用 reindex 方法重置行索引后的 DataFrame:") print(df_reindexed) ``` 原始 DataFrame: A B C 0 1 10 100 1 2 20 200 2 3 30 300 3 4 40 400 4 5 50 500 使用 reindex 方法重置行索引后的 DataFrame: A B C b NaN NaN NaN c NaN NaN NaN d NaN NaN NaN e NaN NaN NaN f NaN NaN NaN