[英]Removing min, max and calculating average
I have columns of numbers and I would need to remove only one min.我有数字列,我只需要删除一分钟。 and one max.
和一个最大。 and then calculate the average of the numbers that remain.
然后计算剩余数字的平均值。 The hitch is that the min/max could be anywhere in the column and some rows may be blank (null) or have a zero, or the column might have only 3 values.
问题是最小值/最大值可能位于列中的任何位置,并且某些行可能为空白(空)或具有零,或者该列可能只有 3 个值。 All numbers will be between 0 and 100. For example:
所有数字都在 0 到 100 之间。例如:
Value Property
80 H
30.5 D
40 A
30.5 A
72 H
56 D
64.2 H
If there is more than one min or max, only one can be removed.如果有多个 min 或 max,则只能删除一个。
To calculate the minimum and maximum of a column, I did as follows:为了计算一列的最小值和最大值,我做了如下操作:
maximum = df['Value'].max()
minimum = df['Value'].min()
In the condition for calculating this average, I also included the condition where it is not null and where it is not equal to zero.在计算这个平均值的条件中,我还包括了它不是 null 和不等于 0 的条件。 However, I do not know how to remove only one max and one min, and add information on greater than 3 rows/values.
但是,我不知道如何只删除一个最大值和一个最小值,并添加关于大于 3 行/值的信息。
I hope you can provide some help/tips on this.我希望你能提供一些帮助/提示。
Let us do idxmin
and idxmax
让我们做
idxmin
和idxmax
out = df.drop([df.Value.idxmax(),df.Value.idxmin()])
Out[27]:
Value Property
2 40.0 A
3 30.5 A
4 72.0 H
5 56.0 D
6 64.2 H
If the objective is to calculate the average without one min and one max, you can just do如果目标是计算没有最小值和最大值的平均值,您可以这样做
(df['Value'].sum() - df['Value'].min() - df['Value'].max())/(len(df)-2)
which outputs 52.54
for your data.它为您的数据输出
52.54
。 Note that this will ignore NaNs etc. This will not modify your df which, if I read the question right, was not the objective anyway请注意,这将忽略 NaN 等。这不会修改您的 df,如果我正确阅读了问题,那无论如何都不是目标
Lately I struggled a little bit with similar problem.最近我遇到了类似的问题。 Finally I came across on numpy.ma library and found this to be elegant solution.
最后我发现了 numpy.ma 库,发现这是一个很好的解决方案。
import numpy.ma as ma
df['Value'].values
# output -> array([80. , 30.5, 40. , 30.5, 72. , 56. , 64.2])
col_name= 'Value'
ma.masked_outside(df[col_name].values, df[col_name].min()+0.02, df[col_name].max()-0.05)
# output -> masked_array(data=[--, --, 40.0, --, 72.0, 56.0, 64.2],
# mask=[ True, True, False, True, False, False, False],
# fill_value=1e+20
# mean for values without outliers
ma.masked_outside(df[col_name].values, df[col_name].min()+0.02, df[col_name].max()-0.05).mean()
声明:本站的技术帖子网页,遵循CC BY-SA 4.0协议,如果您需要转载,请注明本站网址或者原文地址。任何问题请咨询:yoyou2525@163.com.