安装：




```python
pip install sklearn
```

一维数组聚类：
```python
from sklearn.cluster import KMeans
import numpy as np
x = np.random.random(1000).reshape(-1,1)
km = KMeans(n_clusters=5,max_iter=1000).fit(x)
print(km.cluster_centers_)
y = KMeans(n_clusters=5,max_iter=1000).fit_predict(x) #会得出每个sample属于哪一类
print(y)
```

二维数组聚类：

```python
from sklearn.cluster import KMeans
import numpy as np
x = np.random.random(1000).reshape(-1,2)
km = KMeans(n_clusters=5,max_iter=1000).fit(x)
print(km.cluster_centers_)
y = KMeans(n_clusters=5,max_iter=1000).fit_predict(x)#会得出每个sample属于哪一类
print(y)
```

sklearn scikit-learn k-means 聚类算法 一维数组聚类 python 二维数组聚类

例子：





里面的坐标是WGC0084的。

```python
from geopandas import GeoDataFrame
from shapely.geometry import Polygon
import matplotlib.pyplot as plt
record_area = {'新地类': {0: '裸岩', 1: '乔木林'},
               '面积': {0: 2.46, 1: 0.12},
               'Shape_Length': {0: 0.010351519392398598, 1: 0.0023489068258685593},
               'Shape_Area': {0: 2.634822326382532e-06, 1: 1.246329599286182e-07},
               'geometry': {0: Polygon([(115.30992347400002, 41.03045785900008), (115.3098257480001, 41.03042899800005),
                                (115.30969286300001, 41.030391199000064), (115.30960737200007, 41.030363582000064),
                                (115.30965327800004, 41.03029873600008), (115.30983235100007, 41.03028598000003),
                                (115.31004008700006, 41.030290766000064), (115.31030859000009, 41.03026627100007),
                                (115.31056930500006, 41.03019715200003), (115.31077333600001, 41.03014305800008),
                                (115.31089930700011, 41.03007015400004), (115.31105807100005, 41.03001293300008),
                                (115.31135016400003, 41.029988169000035), (115.3116654150001, 41.029762223000034),
                                (115.3118061460001, 41.029661358000055), (115.31249887400008, 41.029374731000075),
                                (115.31273030900002, 41.029320978000044), (115.3130813470001, 41.029239447000066),
                                (115.3131580270001, 41.02922081600008), (115.313297379, 41.029186956000046),
                                (115.31355647200007, 41.02892670500006), (115.31357707600012, 41.02886651400007),
                                (115.31291644200007, 41.02877125500004), (115.31159314900003, 41.028738934000046),
                                (115.31138943700012, 41.02892181000004), (115.31061582600012, 41.02947425800005),
                                (115.30987346500001, 41.029959062000046), (115.30967965600007, 41.030048538000074),
                                (115.30934632200001, 41.03028793000004), (115.30921034800008, 41.03033235700008),
                                (115.3091987890001, 41.03033314600003), (115.30982119300006, 41.03049919700004),
                                (115.30992347400002, 41.03045785900008)]),
                            1: Polygon([(115.30751786600001, 41.03123731200003), (115.30736804700007, 41.03119306200006),
                                (115.30724995300011, 41.03118725400003), (115.30725101300004, 41.03124084600006),
                                (115.30736995600012, 41.03128952700007), (115.30771220700001, 41.03141429200008),
                                (115.30786805900004, 41.03147224600008), (115.30819713300002, 41.03159461300004),
                                (115.30831718700006, 41.03146102800008), (115.3081085990001, 41.03141336600004),
                                (115.30803719100004, 41.03138201500008), (115.3077433850001, 41.031301418000055),
                                (115.30762438200009, 41.03126877200003), (115.30751786600001, 41.03123731200003)])
                            }
               }
g = GeoDataFrame(record_area, crs="EPSG:4326")
print("地图")
print(g)
g.plot()
plt.show()
print("面积直接算 算的图形面积  不是地理面积：")
print(g.area)
g.to_crs(epsg=3310, inplace=True)
print("计算地图面积：")
print(g.area)
```

geopandas 计算面积 crs=“EPSG:4326“转成 3110后计算面积 还是比较准确的 不知道为什么

shapely介绍：





[https://blog.csdn.net/wiborgite/article/details/85167397](https://blog.csdn.net/wiborgite/article/details/85167397)

geopandas介绍：

[https://developer.ibm.com/zh/technologies/analytics/tutorials/working-with-geospatial-vector-data-in-python/](https://developer.ibm.com/zh/technologies/analytics/tutorials/working-with-geospatial-vector-data-in-python/)

官方资料：[https://geopandas.org/aggregation_with_dissolve.html](https://geopandas.org/aggregation_with_dissolve.html)

![在这里插入图片描述](/static/img/images/00753.png)

![在这里插入图片描述](/static/img/images/00754.png)

有一个geopandas的GeoDataFrame对象后：

```python
g['all']=1
qq=g.dissolve(by="all",aggfunc="sum")
qq.plot()
plt.show()
f=qq.loc[1,"geometry"] #拿出图形
fg={"bianjie":list(f.exterior.coords)}  #外环边界
print(list(f.interiors))  #内环
```

获取地图块边界 融合dissolve geopandas

**将python字典写入json文件：





这句就好用了。dumps可以把字典转成字符串，然后文件方法write写入这个字符串。**
```python
with open('jsonFile.json', 'w', encoding="utf-8") as f:
    f.write(json.dumps(zidian, ensure_ascii=False, indent=1))
```

**读取json文件 到python变量（可能是字典或者列表）：**

```python
with open('jsonFile.json', 'r', encoding="utf-8") as f:
    zidian=json.load(f)
```


 load、dump是针对文件的。

 

 loads、dumps是针对字符串的。

json python 读写json 读取和写入json文件 load dump

linux上别人装的环境不知道位置怎么办？





第一步 命令行激活环境

conda activate 环境

第二步 命令行输入python进入写python界面

第三步 随便import一个模块

```python
import os
```

第四步 输出模块的目录

```python
print(os.__file__)
```

第五步 对照着找即可

anaconda 安装位置 linux

Python





```python
l1 = ['b','c','d','b','c','a','a']
l2 = []
[l2.append(i) for i in l1 if not i in l2]
print l2
```

python 列表去重 保持顺序

用这个：





g.to_csv("nihao.csv",encoding="utf_8_sig")

pandas to_csv 中文乱码 解决

http://esri.github.io/geometry-api-java/doc/RelationalOperators.html





![在这里插入图片描述](/static/img/images/00756.png)

![在这里插入图片描述](/static/img/images/00757.png)

![在这里插入图片描述](/static/img/images/00758.png)

![在这里插入图片描述](/static/img/images/00759.png)

![在这里插入图片描述](/static/img/images/00760.png)

![在这里插入图片描述](/static/img/images/00761.png)


![在这里插入图片描述](/static/img/images/00762.png)

![在这里插入图片描述](/static/img/images/00763.png)

![在这里插入图片描述](/static/img/images/00764.png)

python shaply 几何图形关系 Contains Crosses Disjoint Equals Intersects Overlaps Touches Within

极坐标多边形 https://blog.csdn.net/qq_30638831/article/details/79938841





极坐标转直角坐标 https://blog.csdn.net/qq_32425195/article/details/105233463

1。直角坐标转极坐标

import cmath

cn = complex(3,4)

cmath.polar(cn)  #返回长度和弧度


2.极坐标转直角坐标

cn1 = cmath.rect(2, cmath.pi)

cn1.real，cn1.imag





```python
import geopandas
from geopandas import GeoSeries
from shapely.geometry import Polygon
import matplotlib.pyplot as plt
def draw_a_poly(quyu):
    """
    用geopandas画一个多边形
    :param quyu: 二维list
    :return: 无
    """
    p = Polygon(quyu)
    g = GeoSeries(p)
    g.plot()
    plt.show()
import cmath
def get_poly(rad, num_p):
    """
    返回规则num_p边行的各个点
    :param rad: 几何中心到顶点的距离
    :param num_p: 有多少边
    :return: 二维list  N*2
    """
    r = [rad for i in range(num_p)]
    theta = [cmath.pi / num_p * 2 * i for i in range(num_p)]
    ploy = [[cmath.rect(*i).real, cmath.rect(*i).imag] for i in zip(r, theta)]
    return ploy
draw_a_poly(get_poly(2, 16))
```

![在这里插入图片描述](/static/img/images/00765.png)

python 极坐标 多边形点 画多边形 规则多边形 取圆上的若干点

python的logging 不是进程安全的。





第三方模块loguru可以支持进程安全。

注意：不要实例化logger，只需要from loguru import logger即可！

安装：

```python
pip install loguru
```

使用：
```python
from loguru import logger
import datetime
if __name__ == '__main__':
    log_file_name = './log_record/log.log'  # linux路径 我给的相对路径
    logger.add(log_file_name,  # 可以带有路径 没路径的logger会自己创建
               rotation='100 KB',  # 按文件大小切割日志
               enqueue=True,  # 这使得进程安全
               colorize=True,  # 彩色显示 只要系统支持
               level='WARNING')  # 这个级别以上的才会被写入文件，包含这个级别的
    ziduan = {i: 123432 for i in range(10)}
    for i in range(1000):
        logger.debug(ziduan)
        logger.info(ziduan)
        logger.warning(ziduan)
        logger.error(ziduan)
        logger.critical(ziduan)
        logger.critical(datetime.datetime)
```

教程：

[https://huangkai1008.github.io/post/python/loguru/](https://huangkai1008.github.io/post/python/loguru/)

[https://loguru.readthedocs.io/en/stable/api/logger.html](https://loguru.readthedocs.io/en/stable/api/logger.html)

github：

[https://github.com/Delgan/loguru#asynchronous-thread-safe-multiprocess-safe](https://github.com/Delgan/loguru#asynchronous-thread-safe-multiprocess-safe)

效果：

![在这里插入图片描述](/static/img/images/00768.png)

为了让loguru不输出数据到控制台，可以remove掉之前add的东西。

```python
from loguru import logger
import datetime
if __name__ == '__main__':
    log_file_name = './log_record/log__{time}.log'  # linux路径 我给的相对路径
    logger.remove(handler_id=None)  # 清除logger之前的句柄
    logger.add(sink=log_file_name,  # 可以带有路径 没路径的logger会自己创建
               rotation='10 KB',  # 按文件大小切割日志
               retention='30 days',  # 只保留30天的日志
               encoding='utf-8',  # 编码
               enqueue=True,  # 这使得进程安全
               colorize=False,  # 彩色显示 只要系统支持
               format='<green>{time:YYYY-MM-DD HH:mm:ss.SSS}</green> | <level>{level: <8}</level> | <cyan>{name}</cyan>:<cyan>{function}</cyan>:<cyan>{line}</cyan> - <level>{message}</level>',
               level='DEBUG')  # 这个级别以上的才会被写入文件，包含这个级别的
    ziduan = {i: 123432 for i in range(10)}
    for i in range(1000):
        logger.debug(ziduan)
        logger.info(ziduan)
        logger.warning(ziduan)
        logger.error(ziduan)
        logger.critical(ziduan)
        logger.critical(datetime.datetime)
```