Python 添加到已保存CSV的其他列

我有以下从csv生成功能的代码

def gen_features_per_id(file_name, label):
    df = pd.read_csv(file_name, delimiter=',')
    df['dt'] = pd.to_datetime(df['datetime'], unit='s')

    row = []
    column_names = ['group_timestamp', 'label',
                        'x_mean', 'x_median', 'x_stdev', 'x_raw_min', 'x_raw_max', 'x_abs_min', 'x_abs_max',
                        'y_mean', 'y_median', 'y_stdev', 'y_raw_min', 'y_raw_max', 'y_abs_min', 'y_abs_max',
                        'z_mean', 'z_median', 'z_stdev', 'z_raw_min', 'z_raw_max', 'z_abs_min', 'z_abs_max' ]

    group_df = pd.DataFrame(columns=column_names)

    for group_name, g in df.groupby(pd.Grouper(freq='10s', key='dt')):
        print(f'Start time {group_name} has {len(g)} records within 10 secs')

        group_timestamp = group_name
        label = label

        x = g['x'].head(50)
        x_mean = x.mean()
        x_median = x.median()
        x_std_dev = statistics.stdev(x)
        x_raw_min = min(x)
        x_raw_max = max(x)
        x_abs_min = min(abs(x))
        x_abs_max = max(abs(x))

        # print(
        #     f'Mean : {x_mean}, Median : {x_median}, Stdev : {x_std_dev}, '
        #     f'X raw Min : {x_raw_min}, X raw Max : {x_raw_max}, '
        #     f'X abs Min : {x_abs_min}, X abs Max : {x_abs_max}'
        # )

        y = g['y'].head(50)
        y_mean = y.mean()
        y_median = y.median()
        y_std_dev = statistics.stdev(y)
        y_raw_min = min(y)
        y_raw_max = max(y)
        y_abs_min = min(abs(y))
        y_abs_max = max(abs(y))

        # print(
        #     f'Mean : {y_mean}, Median : {y_median}, Std dev : {y_std_dev}, '
        #     f'X raw Min : {y_raw_min}, X raw Max : {y_raw_max}, '
        #     f'X abs Min : {y_abs_min}, X abs Max : {y_abs_max}'
        # )

        z = g['z'].head(50)
        z_mean = z.mean()
        z_median = z.median()
        z_std_dev = statistics.stdev(z)
        z_raw_min = min(z)
        z_raw_max = max(z)
        z_abs_min = min(abs(z))
        z_abs_max = max(abs(z))

        # print(
        #     f'Mean : {z_mean}, Median : {z_median}, Std dev : {z_std_dev}, '
        #     f'X raw Min : {z_raw_min}, X raw Max : {z_raw_max}, '
        #     f'X abs Min : {z_abs_min}, X abs Max : {z_abs_max}'
        # )
        row.append(group_timestamp)
        row.append(label)

        row.append(x_mean)
        row.append(x_median)
        row.append(x_std_dev)
        row.append(x_raw_min)
        row.append(x_raw_max)
        row.append(x_abs_min)
        row.append(x_abs_max)

        row.append(y_mean)
        row.append(y_median)
        row.append(y_std_dev)
        row.append(y_raw_min)
        row.append(y_raw_max)
        row.append(y_abs_min)
        row.append(y_abs_max)

        row.append(z_mean)
        row.append(z_median)
        row.append(z_std_dev)
        row.append(z_raw_min)
        row.append(z_raw_max)
        row.append(z_abs_min)
        row.append(z_abs_max)


        group_df = group_df.append([row], ignore_index=True)
        group_df.to_csv("some.csv", index=False)
        row = []

但保存的csv文件在csv头的开头添加了其他列，这些列等于提供的列数

示例CSV

datetime,x,y,z,label
1493740845,0.0004,-0.0001,0.0045,bad
1493740846,0.0003,0.0002,0.0047,bad
1493740847,0.0005,0.0001,0.0049,bad
1493740848,0.0006,0.0004,0.005,bad
1493740849,0.0006,-0.0003,0.005,bad
1493740851,0.0001,-0.0003,0.0039,bad
1493740852,-0.0006,0.0003,0.0046,bad
B1493740853,0.0001,0.0,0.0048,bad

输出：

0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,group_timestamp,label,x_abs_max,x_abs_min,x_mean,x_median,x_raw_max,x_raw_min,x_stdev,y_abs_max,y_abs_min,y_mean,y_median,y_raw_max,y_raw_min,y_stdev,z_abs_max,z_abs_min,z_mean,z_median,z_raw_max,z_raw_min,z_stdev
# data ... ,,,,,,,,,,,,,,,,,,,,,,,
# data ... ,,,,,,,,,,,,,,,,,,,,,,,

如何解决这个问题

---

---

另外：如果您可以帮助我进一步简化代码。

groupby中的每个循环都有问题，需要将值附加到

行

列表，然后将值附加到

行

外部循环中，以便在上一步中传递到

数据帧

协同构造函数：

#added for nested lists (outside loops)
rows = []

df['dt'] = pd.to_datetime(df['datetime'], unit='s')

for group_name, g in df.groupby(pd.Grouper(freq='10s', key='dt')):
        #added for row per loop
        row = []
        print(f'Start time {group_name} has {len(g)} records within 10 secs')

        group_timestamp = group_name
        label = label 

        x = g['x'].head(50)
        x_mean = x.mean()
        ....
        row.append(z_abs_max)

        rows.append(row)

#DataFrame outside loops
group_df = pd.DataFrame(rows, columns=column_names)
print (group_df)

应通过以下方式改进您的解决方案：

---

您能给出一行样本数据，以便我们执行code@ShubhamPeriwal示例csv输入addedHank它起作用了，我想知道我的方法有什么问题？我假设循环中的数据帧导致了issue@user158-行变量中还追加了值，因此对于eah循环，添加了新的23个值，因此在secon din循环中为46，然后46+23…@user158-hmmm，我很想念它。列具有

1,2,3

的原因是

group\u df=group\u df.append（[row]，ignore\u index=True）

-添加了具有

0,1,2

索引的系列并创建新列。因此应更改为

group\u df=group\u df.append（pd.Series（行，索引=列\u名称），ignore\u index=True）

以防止it@user158-我认为

group\u timestamp

应该在

DatetimeIndex

中，因此在我的解决方案之后应该工作

group\u df=group\u df.reset\u index（）

，并且对于标签使用

group\u df.insert（0，“标签”，标签）

在解决方案之后

#custom aggregate functions
def std_dev(x):
    return statistics.stdev(x)

def abs_min(x): 
    return x.abs().min()

def abs_max(x): 
    return x.abs().max()

d = ['mean','median',std_dev, 'min','max', abs_min, abs_max]
cols = ['x','y','z']

#filtered first 50 rows
df[cols] = df.groupby(pd.Grouper(freq='10s', key='dt'))[cols].head(50)

#aggregate functions
group_df = df.groupby(pd.Grouper(freq='10s', key='dt'))[cols].agg(d)
group_df.columns = group_df.columns.map('_'.join)
print (group_df)