Question

这个问题的标题可能不合适......

因此，假设我有以下input.csv：

Division,id,name
1,3870,name1
1,4537,name2
1,5690,name3

我需要根据id行进行一些处理，取样如下：

>>> get_data(3870)
[{"matchId": 42, comment: "Awesome match"}, {"matchId": 43, comment: "StackOverflow is quite good"}]

我的目标是输出一个csv，它是第一个之间的连接，以及通过get_data检索到的相关数据：

Division,id,name,matchId,comment
1,3870,name1,42,Awesome match
1,3870,name1,43,StackOverflow is quite good
1,4537,name2,90,Random value
1,4537,name2,91,Still a random value
1,5690,name3,10,Guess what it is
1,5690,name3,11,A random value

但是，由于某些原因，在整个过程中，整数数据被转换为float：

Division,id,name,matchId,comment
1.0,3870.0,name1,42.0,Awesome match
1.0,3870.0,name1,43.0,StackOverflow is quite good
1.0,4537.0,name2,90.0,Random value
1.0,4537.0,name2,91.0,Still a random value
1.0,5690.0,name3,10.0,Guess what it is
1.0,5690.0,name3,11.0,A random value

这是我的代码的简短版本，我想我错过了一些东西......

input_df = pd.read_csv(INPUT_FILE)
output_df = pd.DataFrame()

for index, row in input_df.iterrows():
    matches = get_data(row)

    rdict = dict(row)
    for m in matches:
        m.update(rdict)

    output_df = output_df.append(m, ignore_index=True)

    # FIXME: this was an attempt to solve the problem
    output_df["id"] = output_df["id"].astype(int)
    output_df["matchId"] = output_df["matchId"].astype(int)

    output_df.to_csv(OUTPUT_FILE, index=False)

如何将每个浮点列转换为整数？

Answer 1

第一个解决方案是将参数float_format='%.0f'添加到to_csv：

print output_df.to_csv(index=False, float_format='%.0f')
Division,comment,id,matchId,name
1,StackOverflow is quite good,3870,43,name1
1,StackOverflow is quite good,4537,43,name2
1,StackOverflow is quite good,5690,43,name3

第二种可能的解决方案是apply函数convert_to_int而不是astype：

print output_df
   Division                      comment    id  matchId   name
0         1  StackOverflow is quite good  3870       43  name1
1         1  StackOverflow is quite good  4537       43  name2
2         1  StackOverflow is quite good  5690       43  name3

print output_df.dtypes
Division    float64
comment      object
id          float64
matchId     float64
name         object
dtype: object

def convert_to_int(x):
    try:
        return x.astype(int)
    except:
        return x

output_df = output_df.apply(convert_to_int)

print output_df
   Division                      comment    id  matchId   name
0         1  StackOverflow is quite good  3870       43  name1
1         1  StackOverflow is quite good  4537       43  name2
2         1  StackOverflow is quite good  5690       43  name3

print output_df.dtypes
Division     int32
comment     object
id           int32
matchId      int32
name        object
dtype: object

写一个只包含整数值的数据帧

1 个答案: