为什么我在 Python 的这个数据过程中得到 NaN 值？

Question

我试图从 CSV 中获取这四位艺术家各自特征的平均值，但我正在获取这些 NaN 值。

恕我直言，我认为这是因为在 CSV 中有很多零，但后来我尝试从“Extremoduro”中绘制“声学”频率，然后我得到了值的图。

我尝试将平均值放入列中，但它仍然是 NaN 值。另外，我尝试将列格式更改为 int，但也没有任何反应。

这是我的代码：

import numpy as np

result=[]
audio_features=['danceability','energy','key','loudness','mode','speechiness','acousticness','instrumentalness',
            'liveness','valence','tempo','duration_ms']
 artists=["Metallica", "Extremoduro", "AC/DC", "Hans Zimmer"]
 for a in artists:
        for v in audio_features:
                result.append(np.nanmean(df[v].loc[df['name_artist'] == a]))

输出：

                  Metallica  Extremoduro       AC/DC    Hans Zimmer
 danceability       0.349569     0.846328    5.425641      -7.707323
 energy             0.581538     0.098277    0.082615       0.280533
 key                0.413364     0.317938  122.677641  333602.843077
 loudness           0.409805     0.794935    5.857143      -7.786104
 mode               0.571429     0.084675    0.148247       0.169221
 speechiness        0.277273     0.483052  139.082468  257855.370130
 acousticness            NaN          NaN         NaN            NaN
 instrumentalness        NaN          NaN         NaN            NaN
 liveness                NaN          NaN         NaN            NaN
 valence            0.282199     0.269810    4.561889     -18.791699
 tempo              0.593920     0.046895    0.579343       0.742351
 duration_ms        0.159604     0.139642  107.293903  245953.425081

那么，如果我这样做：

import numpy as np

result=[]
audio_features=['danceability','energy','key','loudness','mode','speechiness','acousticness','instrumentalness','liveness','valence','tempo','duration_ms']
 artists=["Metallica", "Extremoduro", "AC/DC", "Hans Zimmer"]
 for a in artists:
        for v in audio_features:
                result.append(np.nanmean(df[v].loc[df['name_artist'] == a]))

输出：

              Metallica  Extremoduro   AC/DC  Hans Zimmer
danceability       0.35         0.85    5.43        -7.71
energy             0.58         0.10    0.32       122.68
key           333602.84         0.41    0.79         5.86
loudness          -7.79         0.57    0.08         0.48
mode             139.08    257855.37     NaN          NaN
speechiness         NaN          NaN     NaN          NaN
valence             NaN          NaN     NaN         0.28
tempo              0.27         4.56  -18.79         0.59
duration_ms        0.05         0.14  107.29    245953.43

另一方面，如果我尝试在我的代码中执行此操作，它实际上会返回一个浮点值：

输入：

a=df['acousticness'].loc[df['name_artist'] == "Metallica"].mean()

输出： 0.08261538461538463

这是我的完整代码：

import matplotlib.pyplot as plt
import pandas as pd
import numpy as np


dataset=pd.read_csv('df.csv')

result=[]
audio_features=['danceability','energy','key','loudness','mode','speechiness','acousticness','instrumentalness',
        'liveness','valence','tempo','duration_ms']
artists=["Metallica", "Extremoduro", "AC/DC", "Hans Zimmer"]
for a in artists:
    for v in audio_features:
        result.append(df[v].loc[df['name_artist'] == a].mean())

result=np.reshape(result,(len(audio_features),len(artists)))
dataset=pd.DataFrame(result,audio_features,artists).round(2)
print(dataset)

Answer 1

源数据可能丢失，NaN 或不正确的 DataType。您可以尝试修复输入csv 文件，或使用fillna 将任何NaN 值填充为0。

df = df.fillna(0)