Pandas中从对象类型列提取数值并进行描述性统计分析

import pandas as pd
import numpy as np

# 假设这是从 'data.xlsx' 加载的数据集结构
# dataset = pd.read_excel('data.xlsx')

# 为演示目的创建一个模拟DataFrame
data = {
    'Memory Speed': ['2666 MHz', '3200 MHz', '1066 MHz', '3200 MHz', 'NaN', '6400 MHz'],
    'Device Weight': ['2 - 4 kg', '1 - 2 kg', 'NaN', '1 - 2 kg', '3 kg', '0.5 kg'],
    'Screen Size': ['15.6 inches', '15.6 inches', '10 inches', '15.6 inches', '14 inches', '17.3 inches'],
    'GPU Memory Size': ['Shared', '1 GB', '2 GB', 'Shared', '4 GB', '8 GB'],
    'SSD Capacity': ['256 GB', '512 GB', '1 TB', '1 TB', '2 TB', '500 GB'],
    'Max Processor Speed': ['3.4 GHz', '3.6 GHz', '1.05 GHz', '4.2 GHz', '5.0 GHz', '4.8 GHz'],
    'RAM (System Memory)': ['4 GB', '8 GB', '12 GB', '16 GB', '32 GB', '64 GB'],
    'Backlit Keyboard': [0, 1, 0, 0, 1, 1],
    'Price': ['Very Low', 'Low', 'High', 'Medium', 'Low', 'Medium']
}
dataset = pd.DataFrame(data)

print("原始数据集信息:")
dataset.info()
print("\n原始数据集头部:")
print(dataset.head())

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 6 entries, 0 to 5
Data columns (total 9 columns):
 #   Column               Non-Null Count  Dtype 
---  ------               --------------  ----- 
 0   Memory Speed         6 non-null      object
 1   Device Weight        5 non-null      object
 2   Screen Size          6 non-null      object
 3   GPU Memory Size      6 non-null      object
 4   SSD Capacity         6 non-null      object
 5   Max Processor Speed  6 non-null      object
 6   RAM (System Memory)  6 non-null      object
 7   Backlit Keyboard     6 non-null      int64 
 8   Price                6 non-null      object
dtypes: int64(1), object(8)
memory usage: 592.0+ bytes

import pandas as pd
import numpy as np

# 假设 df 是从 Excel 加载的 DataFrame
# df = pd.read_excel("Dataset.xlsx") 

# 为演示目的，使用前面定义的模拟数据集
df = dataset.copy() 

df_new = pd.DataFrame(columns=df.columns) # 创建一个空的新DataFrame，保留列名

for col in df.columns: # 遍历原始DataFrame的每一列
    if df[col].dtype == "object": # 检查列是否为对象类型
        values = [] # 初始化一个空列表来存储转换后的值
        for val in df[col].values: # 遍历当前列的每一个值
            if pd.isna(val) or (isinstance(val, str) and val.upper() == 'NAN'): # 处理NaN值（包括字符串'NaN'）
                values.append(np.nan)
            elif isinstance(val, str): # 确保值是字符串才进行解析
                val_trimmed = val.strip() # 去除首尾空白

                # 处理包含空格的字符串，通常表示“数值 单位”
                if " " in val_trimmed:
                    val_parts = val_trimmed.split(" ")
                    # 尝试转换第一个部分为数值
                    try:
                        # 替换逗号为小数点，以处理不同地区的小数表示
                        numeric_part = val_parts[0].replace(",", ".")
                        num = float(numeric_part)

                        # 特殊处理SSD Capacity列的单位转换 (GB to TB)
                        if col == "SSD Capacity" and len(val_parts) > 1:
                            unit = val_parts[1].upper()
                            if unit == "GB":
                                values.append(num / 1000) # 转换为TB
                            elif unit == "TB":
                                values.append(num)
                            else: # 未知单位，保留原值
                                values.append(val)
                        # 处理RAM (System Memory)列的单位转换 (GB)
                        elif col == "RAM (System Memory)" and len(val_parts) > 1:
                            unit = val_parts[1].upper()
                            if unit == "GB":
                                values.append(num) # 已经是GB，直接添加
                            else: # 未知单位，保留原值
                                values.append(val)