算法数据分析实例——城市气候与海洋关系数据分析[线性回归y=w*x+b]

城市气候与海洋的关系研究

导入包

import numpy as np

import pandas as pd
from pandas import Series,DataFrame

import matplotlib.pyplot as plt
%matplotlib inline

城市信息

citis = np.array(['asti','bologna','cesena','faenza','ferrara','mantova','milano','piacenza','ravenna','torino'])

数据集如下图：

 

选择某个城市进行查看

df1 = pd.read_csv('asti_150715.csv')
df1

df2 = pd.read_csv('asti_250715.csv')
df2

df3 = pd.read_csv('asti_270615.csv')
df3

# pandas进行级联的时候，行索引是不是可以重复
df = pd.concat([df1,df2,df3],ignore_index=True)
df

df.drop('Unnamed: 0',axis = 1)

使用程序快速导入数据集

citis = np.array(['asti','bologna','cesena','faenza','ferrara','mantova','milano',
'piacenza','ravenna','torino'])

citis.size
Out： 10

# 使用城市名称作为键，城市数据为值 存储至字典
data = {}
for city in citis:
    df1 = pd.read_csv('%s_150715.csv'%(city))
    df2 = pd.read_csv('%s_250715.csv'%(city))
    df3 = pd.read_csv('%s_270615.csv'%(city))
    
    # 合并
    df = pd.concat([df1,df2,df3],ignore_index=True)
    
    df.drop('Unnamed: 0',axis = 1,inplace=True)
    
    data[city] = df

 

各城市与海洋距离，最高温度，最低温度，最高湿度，最低湿度

data['milano']

info = []
for city in citis:
    df = data[city]
    
    temp_min = df['temp'].min()
    temp_max = df['temp'].max()
    
    humidity_min = df['humidity'].min()
    humidity_max = df['humidity'].max()
    
    dist = df['dist'][0]
    
    info.append([city,temp_min,temp_max,humidity_min,humidity_max,dist])

使用选取的数据新建一个数据表

df_cities = DataFrame(info,columns=['city','temp_min','temp_max','humidity_min',
                                    'humidity_max','dist'])
df_cities

显示最高温度与离海远近的关系

df_cities.plot('dist','temp_max',kind = 'scatter')

 

线性回归：算法解决问题

# 算法解决问题
# 线性回归，研究一些数据之后，规律归纳出来------> 方程
from sklearn.linear_model import LinearRegression

lrg = LinearRegression()

 

近海和最高温度关系

# 求解近海的数据 < 150

cond = df_cities['dist'] < 150

# 训练数据，有格式要求：必须是二维的
[[样本一],[样本二],[样本三]……]
X = df_cities[['dist']][cond]

y = df_cities['temp_max'][cond]

# 训练，近海的数据和海洋的距离，目标值，最高温度
lrg.fit(X,y)
Out： LinearRegression(copy_X=True, fit_intercept=True, n_jobs=1, normalize=False)

# 算法可以接收ndarray
X1 = X.values
y1 = y.values
lrg.fit(X1,y1)
Out： LinearRegression(copy_X=True, fit_intercept=True, n_jobs=1, normalize=False)

计算y=w*x+b

w_ = lrg.coef_[0]

b_ = lrg.intercept_

display(w_,b_)
Out: 
0.013760226316996685
32.61657542625584

# f(x) = w*x + b

plt.scatter(df_cities['dist'],df_cities['temp_max'])


# 绘制机器学习获得线
x = np.linspace(0,150,10)

# 趋势
y = w_*x + b_

plt.plot(x,y,color = 'green')

远海和最高温度关系

# 获取远海的数据 >120 
cond = df_cities['dist'] > 120

X = df_cities[['dist']][cond]

y = df_cities['temp_max'][cond]

lrg = LinearRegression()

lrg.fit(X,y)
Out： LinearRegression(copy_X=True, fit_intercept=True, n_jobs=1, normalize=False)

# weight 权重
w1_ = lrg.coef_[0]

# bias 偏差 f(x) = w*x + b 
b1_ = lrg.intercept_
display(w1_,b1_)

# 机器学习，人工智能
# 数学问题

计算y=w*x+b

# f(x) = w*x + b

plt.scatter(df_cities['dist'],df_cities['temp_max'])


# 绘制机器学习获得线
x = np.linspace(0,150,10)

# 趋势
y = w_*x + b_

plt.plot(x,y,color = 'green')

# 远海
x = np.linspace(120,360,100)

y = w1_*x + b1_

plt.plot(x,y,color = 'red')

 

用numpy创建一个直方图，将360度划分为8个面元，将数据分类到这8个面元中

wind_deg = data['milano']['wind_deg']
wind_deg
Out： 
0     100
1       0
2     140
3       0
4      80
5       0
……

 

wind_data,wind_range = np.histogram(wind_deg,bins=8,range=[0,360])
wind_data
Out：
array([21,  9,  9,  6, 14,  3,  2,  2], dtype=int64)

 

plt.figure(figsize=(8,8))
plt.axes(polar = True)

colors = np.random.rand(8,3)

plt.bar(np.arange(0,2*np.pi,np.pi/4),wind_data,color = colors)