Python数据分析：利用Pandas和Numpy处理SNS数据，计算电导率

import os
import pandas as pd
import numpy as np
from openpyxl import Workbook, load_workbook
from openpyxl.utils.dataframe import dataframe_to_rows

class SNSData:
    def __init__(self, pandas_dir=''):#open 文件夹
        if os.path.isfile(pandas_dir):
            self.path = pandas_dir
        else:
            raise FileNotFoundError('File not found:' + pandas_dir)
        self.raw_data = pd.read_csv(self.path, encoding='utf-8')
        self.regular_data = {'name': [], 'property':[],  'magnitude': [], 'phase': [], 'frequency': []}
        self.freq_list = [1, 2, 3, 7, 11, 17, 23, 31, 43, 61, 89, 127, 179, 251, 349]

        for it in self.freq_list:
            self.regular_data['name'].extend(self.raw_data['name'])
            self.regular_data['property'].extend(self.raw_data['property'])  # 盐水考虑浓度属性，脑脊液不考虑, 在新数据中命名为盐水+浓度
            self.regular_data['magnitude'].extend(self.raw_data[str(it) + 'kHz_mag'])
            self.regular_data['phase'].extend(self.raw_data[str(it) + 'kHz_ang'])
            for i in range(self.raw_data.shape[0]):
                self.regular_data['frequency'].append(it)

        self.regular_data['logFreq'] = np.log10(self.regular_data['frequency'])  # 将频率以对数形式展现
        self.regular_data['R'] = np.array(self.regular_data['magnitude']) * np.cos(
            np.array(self.regular_data['phase']) * 3.14 / 180)  # np.cos默认弧度
        self.regular_data['X'] = np.array(self.regular_data['magnitude']) * np.sin(
            np.array(self.regular_data['phase']) * 3.14 / 180)
        self.regular_data['logMag'] = np.log10(self.regular_data['magnitude'])  # 在bode图中，幅度也是用对数表示的
        '''下面利用已知浓度盐水的R和电导率（conductivity）映射关系来得到待测物体的电导率'''
        # 创建电导率空列表保持行数一致
        self.regular_data['conductivity'] = [0] * self.raw_data.shape[0] * len(self.freq_list)
        self.df = pd.DataFrame(self.regular_data)
        self.generate_conductivity()

        '''重新整理完数据后，整合至pandas数据帧，用get_df方法可以返回该pandas数据帧'''

    def generate_conductivity(self):
        '''
        映射关系如下
        利用给出的盐水和电导率表自己先写个映射关系
        '''
        conductivity_list = [976, 1987, 3850, 5650, 7450, 9238]  # 该电导率列表对应浓度为0.1,0.5的盐水
        concentration_list = [0.05, 0.10, 0.2, 0.3, 0.4, 0.50]
        saline_real_dict = {}  # 每个频率，每个浓度的盐水的实部，其中同浓度对应同一个电导率

        # 在这里，首先过滤出同一个频率下的所有盐水的real part，然后对同一个浓度的realpart求平均值，这里有七个浓度，将七个平均值打包为一个列表，作为子列表添加入saline_real_list
        for it in self.freq_list:  # 对每一个频率
            one_freq_saline_real_list = []  # 打包用的子列表

            for con in concentration_list:
                # 这里依次为0.05， 0.1, 0.2, 0., 0.4, 0.6的盐水的real part平均值
                saline_concentration = 'saline' + format(con, '.2f')
                filted_df = self.df[
                    self.df['frequency'].isin([it]) * self.df['name'].isin([saline_concentration])]  # 筛选出该频率下浓度0.05的样本
                avg = np.mean(filted_df['R'])  # 求筛选出来的下R的平均值
                one_freq_saline_real_list.append(avg)

            saline_real_dict[it] = one_freq_saline_real_list  # 每一个频率下都有了一个浓度对应的real part列表
            # real part和电导率建立起映射关系，对于单个频率，one_freq_saline_real_list作为x轴，conductivity作为y轴，用np.interp方法, 寻找R对应的电导率
            self.df['conductivity'] += self.df['frequency'].isin([it]) * np.interp(self.df['R'],
                                                                                   one_freq_saline_real_list[::-1],
                                                                                   conductivity_list[
                                                                                   ::-1])  # interp要求x轴为递增序列，因此此处倒序输入

    def get_df(self):
        return self.df
    print(os.getcwd())
def save_20230213NO1_conductivity(path):
     # df = pd.DataFrame(regular_data)
    snsdata = SNSData(pandas_dir=path)
    df = snsdata.get_df()
    # 选择特定类型
    filtered_df = df[~df['name'].isin(['air'])]
    rslt_df = filtered_df[filtered_df['property'].str.contains('A|B|C|D|E|F|G|H') & filtered_df['name']. str.contains('saline|normal|tumor|peritumor')]
    wb = Workbook()
    ws = wb.active
    for r in dataframe_to_rows(rslt_df, index=True, header=True):
        ws.append(r)
    wb.save('2023_2_13No1_data1.csv')
    return None


# snsdata = SNSData()
# snsdata.save_20230213NO1_conductivity()
def calculate_conductivity(path):
    df = pd.read_csv( '2023_2_13No1_data1.csv')
    index = [1, 2, 3, 7, 11, 17, 23, 31, 43, 61, 89, 127, 179, 251, 349]
    my_df = []
    for i in range(len(index)):
        the_df = df[(df['name'] == 'tumor') & (df['property'] == 'A|B|C|D|E|F|G|H') & (df['frequency'] == index[i]) & (
                    df['conductivity'] > 0)]
        my_df.append(the_df)

    result = []
    for item in my_df:
        result.append((sum(item['conductivity'])) / (item.shape[0]))

    df_result = pd.DataFrame({'result': result})
    df_result.to_csv('2023_2_13No1_data2.csv', index=False)
    return None

snsdata = SNSData(pandas_dir='./2023_2_13No1/2023_2_13_12.csv')#jiu'zhe'yi'ge'wen'ti'dui'ba
save_20230213NO1_conductivity('./2023_2_13No1/2023_2_13_12.csv')
calculate_conductivity('./2023_2_13No1/2023_2_13_12.csv')
snsdata = SNSData(pandas_dir='./2023_2_13No1/2023_2_13_12.csv')
df = snsdata.get_df()
df