描述性数据分析-Python实现

描述性数据分析-Python实现

一、集中趋势

• 算术平均值

  data=[1,4,8,10,12]
  #Python 函数实现
  def arithmetic_mean(list):
      '''
          算术平均值:数据集合中的数据之和除以数据值个数
      '''
      list_length = len(list)
      sum = 0
      for i in list:
          sum += i
      return sum / list_length
  
  print(arithmetic_mean(data))
  
  
  #Numpy实现
  import numpy as np
  arithmetic_mean = np.mean(data)
  print(arithmetic_mean)
  
  

• 加权平均值

  data=[1,4,8,10,12]
  weight=[0.1,0.2,0.3,0.4,0.5]
  #Python 函数实现
  def weighted_average(data_list,weight_list):
      '''
          加权平均数：所有数值与权重的乘积之和除以权重和
      '''
      return sum([data_list[i]*weight_list[i] for i in range(len(weight_list))])/sum(weight_list)
  
  print(weighted_average(data,weight))
  
  #Numpy实现
  import numpy as np
  weighted_average = np.average(data,weights=weight)
  print(weighted_average)
  

• 几何平均值

  data=[1,4,8,10,12]
  #Python 函数实现
  def geometric_mean(data_list):
      '''
          几何平均数：对数值乘积开n次方根，n为数值的个数
      '''
      sum = 1 
      for i in data_list:
          sum *=i
      return pow(sum,1.0/len(data_list))
  
  print(geometric_mean(data))
  
  #Scipy实现
  from scipy import stats as sts
  geometric_mean = str(sts.gmean(data))
  print(geometric_mean)
  

• 众数

  data=[1,4,8,10,12,12]
  #Python 函数实现
  def majority_element(nums):
          count = {}
          for num in nums:                                                # 统计每个数字出现的次数
              if num in count:
                  count[num] += 1
              else:
                  count[num] = 1
          return {v: k for k, v in count.items()}[max(count.values())]    # 字典键值反转，找到出现次数最多的数字
  print(majority_element(data))
  
  
  #Numpy实现
  import numpy as np
  counts = np.bincount(data)
  #返回众数
  majority_element = np.argmax(counts)
  print(majority_element)
  

• 中位数

  data=[1,4,8,10,12,12]
  #Python 函数实现
  def median(data_list):
      length = len(data_list)
      data_list.sort()
      if (length % 2) == 1:
          z= length // 2
          y = data_list[z]
      else:
          y = (data_list[length // 2]+data_list[length // 2 - 1]) / 2
      return y
  print(median(data))
  
  #Nump 实现
  import numpy as np
  median = np.median(data)
  print(median)
  

二、离散程度

• 极差

  data=[1,4,8,10,12,12]
  #Python 函数实现
  def very_poor(data_list):
      return max(data_list)-min(data_list)
  
  print(very_poor(data))
  
  #Numpy实现
  import numpy as np
  very_poor = np.ptp(data)
  print(very_poor)
  

• 平均差

  data=[4,8]
  #Python 函数实现
  def average_difference(data):
      '''
          平均差：数据与平均值之间的绝对值之和除以数值个数
      '''
      avg = arithmetic_mean(data)
      sum = 0 
      for i in data:
          sum += abs(i-avg)
      return sum / len(data)
  
  print(average_difference(data))
  
  #Numpy实现
  import numpy as np
  a = np.array(data)
  average_difference = np.sum(abs(a-np.mean(a)))/len(a)
  print(average_difference)
  

• 方差和标准差

  data=[1,4,8,10,12]
  #Python 函数实现
  def variance(data_list):
      #方差
  	avg = arithmetic_mean(data_list)
  	sum = 0
  	for i in data_list:
  		sum += pow(i-avg,2)
  	
  	return sum/len(data_list)
  
  
  def standard_deviation(data_list):
      #标准差
  	return pow(variance(data_list),1/2)
  
  
  print(variance(data))
  print(standard_deviation(data))
  
  
  #Numpy实现
  import numpy  as np
  print(str(np.var(data))) #方差
  print(str(np.std(data))) #标准差
  

• 四分位极差和四分位差以及离散系数

  import numpy  as np
  data=[1,4,8,10,12]
  print(str(np.percentile(data,75))) #四分位 75%
  print(str(np.percentile(data,25))) #四分位 25%
  print(str(np.percentile(data,75) - np.percentile(data,25)))#四分位极差
  print(str(np.std(data)/np.mean(data))) #离散系数