Python常用代码（自己平时用得比较多）

pandas dataframe merge 合并两个df

data = pd.merge(data1, data2, 'left', on=['id'])

 

更改某列数据类型

data['result'] = data['result'].astype(int)
data['result'] = data['result'].astype(float)

取某几列给新的df

result = data[['id', 'result']]

 

更改列名

a.rename(columns={'A':'a', 'B':'b', 'C':'c'}, inplace = True)

 

显示所有列

# 显示所有列
# pd.set_option('display.max_columns', False)
# pd.set_option('display.max_rows', 500)
pd.set_option('display.max_columns', None)
pd.set_option('display.width', 3000)
# 这两个参数的默认设置都是False
pd.set_option('display.unicode.ambiguous_as_wide', True)
pd.set_option('display.unicode.east_asian_width', True)

 

获取第几行第几列

data.ix[1:-1,[3, 4]]

 

遍历行（不能修改数据）

for index, row in df.iterrows():
    print row["c1"], row["c2"]

遍历行并修改数据

def valuation_formula(x, y):
    return x * y * 0.5
 
df['price'] = df.apply(lambda row: valuation_formula(row['x'], row['y']), axis=1)

或者

for index, row in dataset.iterrows():
    a, b, c, d = func(row[:-4])
    dataset.ix[index, ['2018-1']] = int(a)
    dataset.ix[index, ['2018-2']] = int(b)
    dataset.ix[index, ['2018-3']] = int(c)
    dataset.ix[index, ['2018-4']] = int(d)

 

修改满足条件的行

evaluation.loc[(evaluation['province'] == row['province']) & (evaluation['model'] == row['model']) & (
                    evaluation['regMonth'] == 1), 'forecastVolum'] = row['2018-1']

 

某列最大 /小值

df['row'].max()

 

df最大最小值

df.max().max()

查看各列数据类型

df.dtypes

按条件取行

df = df[df['STATUS'] == "ACTIVE"]

两个df合并

result = left.join(right, on='key')

 

将某一列转化成list

df['a'].tolist()

字符串与时间格式相互转化

import datetime
# str转时间格式：
dd = '2019-03-17 11:00:00'
dd = datetime.datetime.strptime(dd, "%Y-%m-%d %H:%M:%S")
print(dd,type(dd))

# 时间格式转str:
dc = dd.strftime("%Y-%m-%d %H:%M:%S")
print(dc,type(dc))

 

创建df

df = pd.DataFrame(columns=[???])

逐行追加，增加一行

df.loc[len(df)] = [16, 17, 18, 19]

 

统计某列元素出现的个数（唯一）

df.value_counts()

 

df排序

log_df = log_df.sort_values(['datetime', 'blockID'])

 

df分组，并进行迭代

for blockID, group in df.groupby('key1'):
        print (blockID)
        print (group)

 

df某列转化成list

# 方法1
df['a'].values.tolist()
 
# 方法2
df['a'].tolist()

# 方法3（不重复元素）
df['a'].drop_duplicates().values.tolist()

 

list去除重复元素，不改变顺序

word_list = sorted(set(word_list), key=word_list.index) 

 

写文件

with open(path + 'log_seq.txt', 'w', encoding='utf-8') as f:
    f.write(log_seq)

获取某列的唯一值

data.drop_duplicates(['name'])['name']

 

将秒转化成时分秒

import time
print(time.strftime("%H:%M:%S", time.gmtime(666)))

 

读txt文件

# 读txt文件
def read_data(path):
    content = []
    file = open(path, 'r', encoding='UTF-8')
    for line in file:
        content.append(line.strip())
    file.close()
    return content

 

读csv文件 

# 读csv文件
def read_data_csv(path):
    birth_data = []
    with open(path, encoding="utf-8") as csvfile:
        csv_reader = csv.reader(csvfile)  # 使用csv.reader读取csvfile中的文件
        birth_header = next(csv_reader)  # 读取第一行每一列的标题
        for row in csv_reader:  # 将csv 文件中的数据保存到birth_data中
            birth_data.append(row)
    return birth_data

 

写入csv文件 

# 写入csv文件
def write_data_csv(path, data):
    with open(path, "a", encoding="utf-8", newline='') as csvfile:
        writer = csv.writer(csvfile)
        # 写入多行用writerows
        temp = data.split(',')
        temp_list = []
        for i in range(len(temp)):
            temp_list.append(temp[i])
        writer.writerows([temp_list])

 

 文本预处理：中文分词，去听用词

# 文本预处理：中文分词，去除停用词

import jieba

def pre_work(string):
    # 分词，去停用词
    words = ' '.join(jieba.cut(string)).split(' ')  # 利用jieba工具进行中文分词

    # 获取停用词
    stopwords = set()
    file = open("stopwords.txt", 'r', encoding='UTF-8')
    for line in file:
        stopwords.add(line.strip())
    file.close()

    text = []
    # 过滤停用词，只保留不属于停用词的词语
    for word in words:
        if word not in stopwords:
            text.append(word)
    return text

 

pandas归一化（放到另外一列）

loss_df['loss_normalization'] = 0
d = loss_df['loss'].apply(lambda x: (x - loss_df['loss'].min()) / (loss_df['loss'].max() - loss_df['loss'].min()))
print(d)
loss_df['loss_normalization'] = d