新冠状病毒预测

闲来无事，做一个确诊人数的预测吧，希望拐点早日降临。
首先数据就是日期和总确诊人数，走势是平缓到爆发到平缓，所以用logistics函数。总治愈目前处于上升趋势，多项式拟合吧。大体思路就是自变量特征从1到总的天数，然后把数字映射到日期，制图。
附上代码

python

import numpy as np
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from scipy.optimize import curve_fit



def date_encode(date):
    # '01.24' -> 1 * 100 + 24 = 124
    d = date.split('/')
    return int(d[1]),int(d[2])

def date_decode(date):
    # 124 -> '01.24'
    return '{}.{}'.format(str(date // 100), str(date % 100))


df = pd.read_csv('data.csv')#encoding='utf-8',header=None,sep = '\t'
df.drop([33],inplace=True)
X = np.array(df.iloc[:,0])    #日期太多显示会重叠，前十天数据很平缓故先忽略掉
cur_month,cur_day=date_encode(X[0])
y = np.array(df['total_confirmed'])
z= np.array(df['new_recoveries'])
x = np.arange(len(y))

def get_date_list(cur_month,cur_day,days,prediction=7):
    """
    得到原始数据和预测的日期
    """
    month_day = [0, 31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31]
    ans = []
    n=days+prediction
    while n:
        if cur_day <= month_day[cur_month]:
            d = "0" + str(cur_day) if cur_day < 10 else str(cur_day)
            ans += [str(cur_month) + "/" + d]
            cur_day += 1
            n=n-1
        else:
            cur_day = 1
            cur_month += 1
            n=n-1
    return ans


ans = get_date_list(cur_month,cur_day,len(y),prediction=7)
def logistic_function(t, K, P0, r):
    r=0.27
    t0 = 0
    exp = np.exp(r * (t - t0))
    return (K * exp * P0) / (K + (exp - 1) * P0)

def f_3(x, A, B, C, D):  
    return A*x*x*x + B*x*x + C*x + D


popt, pcov = curve_fit(logistic_function, x, y)
popt1, pcov1 = curve_fit(f_3, x, z)
predict_x = list(x)+[x[-1] + i for i in range(1, 8)] #数组合并不能直接相加
predict_x = np.array(predict_x)
predict_y = logistic_function(predict_x, popt[0], popt[1], popt[2])
predict_y = [int(i) for i in predict_y]
predict_z = f_3(predict_x, popt1[0], popt1[1], popt1[2],popt1[3])
predict_z = [int(i) for i in predict_z]
#print(ans[-7:],predict_y[-7:])

#输出新增确诊
new_infected = [predict_y[i]-predict_y[i-1] for i in range(-7,0)] 
print(ans[-7:],new_infected)


#plt.scatter(x,y,color='purple',label='real')
#plt.plot(x,y,color='gray')
#plt.scatter(predict_x,predict_y,marker='x',color='red',label='predicted data')
#plt.xticks(predict_x,ans,rotation=90)
#plt.suptitle("Logistic Fitting Curve for 2019-nCov total infected numbers", fontsize=16, fontweight="bold")

#输出新增治愈
new_cured = [predict_z[i]-predict_z[i-1] for i in range(-7,0)] 
print(ans[-7:],new_cured)

plt.scatter(x,z,color='purple',label='real')
plt.plot(x,z,color='gray')
plt.scatter(predict_x,predict_z,marker='x',color='red',label='predicted data')
plt.xticks(predict_x,ans,rotation=90)


plt.suptitle("polynomial regression Fitting Curve for 2019-nCov total cured numbers", fontsize=16, fontweight="bold")
plt.xlabel('date', fontsize=14)
plt.ylabel('infected number', fontsize=14)
plt.show()

预测走势

新增确诊人数：

新增治愈趋势：

新增治愈人数：

导出csv文件

导出文件，list作为列的方法：

python

a = ['2020/2/24', '2020/2/25', '2020/2/26', '2020/2/27', '2020/2/28', '2020/2/29', '2020/3/01']
b = [307, 236, 181, 139, 106, 81, 63]
c = [2818, 3019, 3229, 3447, 3673, 3909, 4153]
data = {'date':a,'new_confirmed':b,'new_recoveries':c}
dataframe = pd.DataFrame(data)
dataframe.to_csv(r'D:\python\test.csv')

结果

用list导入dataframe是list[[行],[行]]

python

list=[new_infected[i] for i in range(-7,0)]
list1=[predict_z[i] for i in range(-7,0)]
list.extend(list1)
list=[list]
column=['confirmed_day1','confirmed_day2','confirmed_day3','confirmed_day4','confirmed_day5','confirmed_day6','confirmed_day7','recovery_day1','recovery_day2','recovery_day3','recovery_day4','recovery_day5','recovery_day6','recovery_day7']  
test=pd.DataFrame(columns=column,data=list,index=['total'])
test.to_csv('D:/test1.csv')