编写一个函数，在 pandas 数据框中连续汇总某些值

Question

我有一个 pandas DataFrame，我想编写一个函数来帮助我将每个负值总结为 result1，并将每个正值总结为 result2。所以基本上，这个函数应该遍历列“total_load”

def total_battery(ok6, col_name='total_load'):
"""Return a dictionary with counts of occurrences."""


# Initialize an empty dictionary: cols_values
cols_values = {}

# Extract column from df: col
col = ok6[col_name]

# iterate over the column in df
for entry in col:


    if entry in cols_values.keys() < 0: ***--> then sum all the negative values*** 
        cols_values[entry] += sum

    else: 
        if entry in cols_values.keys() > 0: ***--> then sum all the negative values*** 
            cols_values[entry] += sum


    # Return the cols_count dictionary
    return cols_values

# Call count_entries(): result1
result1 = total_battery(ok6, "total_load")

# Call count_entries(): result2
result2 = total_battery(ok6, "total_load")

# Print result1 and result2
print(result1)
print(result2)

我希望这不会太令人困惑。

感谢您的帮助。

Answer 1

使用boolean indexing 或query 进行过滤，然后使用Series.sum：

result1 = df.loc[df['total_load'] < 0, 'total_load'].sum()
result2 = df.loc[df['total_load'] > 0, 'total_load'].sum()

---

result1 = df.query('total_load < 0')['total_load'].sum()
result2 = df.query('total_load > 0')['total_load'].sum()

示例：

rng = pd.date_range('2016-06-01', periods=4, freq='T')
df = pd.DataFrame({'total_load':[1,2,-3,-5]}, index=rng)
print (df)
                     total_load
2016-06-01 00:00:00           1
2016-06-01 00:01:00           2
2016-06-01 00:02:00          -3
2016-06-01 00:03:00          -5

result1 = df.loc[df['total_load'] < 0, 'total_load'].sum()
result2 = df.loc[df['total_load'] > 0, 'total_load'].sum()
print (result1)
-8
print (result2)
3

result1 = df.query('total_load < 0')['total_load'].sum()
result2 = df.query('total_load > 0')['total_load'].sum()
print (result1)
-8
print (result2)
3