计算数据帧中数据段的平均值

问题描述

在下面的图片中，我有一个巨大的数据帧。对于每一次冲程，机器都会呈现穿透值，然后给零。我想计算一下每划一划的平均数。例如，[0.762，0.766]的平均值，以及[0.66，1.37，2.11，2.29]的平均值，依此类推，直到数据帧结束。请注意，笔划没有固定的长度。enter image description here

---

解决方案

# Generate example data based on your image
df = pd.DataFrame({'penetration': 
                   [0, 1, 0, 2, 3, 0, 0, 
                    5, 6, 7, 0, 0, 0, 0]})

# Flag rows with nonzero penetration depth
df['segment'] = df['penetration'].ne(0)

# Flag rows which represent a change from non-segment
# to segment, or from segment to non-segment
df['group'] = df['segment'].ne(df['segment'].shift())

# Label each segment with a unique integer
df['group'].cumsum()

# Zero out the group number of rows which belong to non-segments
df['group'] = df['group'].where(df['segment'], 0)

# Get mean of penetration depths for each group
df['mean'] = df.groupby('group')['penetration'].transform('mean')

# Print result
df

    penetration  segment  group  mean
0             0    False      0   0.0
1             1     True      2   1.0
2             0    False      0   0.0
3             2     True      4   2.5
4             3     True      4   2.5
5             0    False      0   0.0
6             0    False      0   0.0
7             5     True      6   6.0
8             6     True      6   6.0
9             7     True      6   6.0
10            0    False      0   0.0
11            0    False      0   0.0
12            0    False      0   0.0
13            0    False      0   0.0