安全库存优化--需求概率分布拟合

# -*- coding: utf-8 -*-

def fit_model(data,dist):
    from scipy import stats
    from fitter import Fitter

    f = Fitter(data,timeout=100,distributions=[dist])
    f.fit()
    return f.get_best()


def run_():
    # 数据生成
    from scipy import stats
    import matplotlib.pyplot as plt

    data = stats.gamma.rvs(2, loc=1.5, scale=2, size=10000)  # 通过scipy生成服从gamma分布的10000个样本
    # 拟合分布
    from fitter import Fitter

    f = Fitter(data, timeout=60)  # 创建Fitter类
    f.fit()  # 调用fit函数拟合分布
    print(f.summary())# 输出拟合结果
    # f.hist()
    plt.show()
    print('最佳分布：',f.get_best())
    # print(f.plot_pdf(Nbest=5, lw=2, method='sumsquare_error'))


def run():
    from pylab import linspace, plot
    import scipy.stats

    dist = scipy.stats.gamma
    param = (1.9870, 1.5026, 2.0174)
    X = linspace(0, 10, 10)
    pdf_fitted = dist.pdf(X, *param)
    plot(X, pdf_fitted, 'o-')



if __name__ == '__main__':
    run_()

# -*- coding: utf-8 -*-

"""偏度计算，判断数据分布是否存在长尾"""

import numpy as np

def sample_skewness(data):
    """样本偏度"""
    mean = np.mean(data)
    mid = np.median(data)
    std = np.std(data)

    if std==0:
        return 0
    skewness = 3*(mean-mid)/std
    return skewness

def third_central_moment(data):
    """三阶中心矩计算"""
    mean = np.mean(data)
    std = np.std(data)
    data_list = [((x-mean)/std)**3 for x in data]
    skewness = np.mean(data_list)
    return skewness

if __name__ == '__main__':
    data = [1,2,3,4,2,1]
    print(sample_skewness(data),third_central_moment(data))

# -*- coding: utf-8 -*-
import numpy as np

from fit import skewness as sk
from fit import prodist as dt

def dist(data,buff):
    '''
    拟合data,得到的最佳的概率分布函数
    :param data: 待拟合分布的数据
    :param buff: 偏度阈值，在偏度范围内拟合高斯分布，之外拟合gamma分布
    :return: 返回概率分布，字典方式
    '''
    skew = sk.sample_skewness(data)
    print("skew:",skew)

    dtb = None
    if np.absolute(skew)>=buff:
        "gammma or 泊松分布 拟合"
        dtb = dt.fit_model(data,"gamma")

    else:
        "高斯分布拟合"
        dtb = dt.fit_model(data,"norm")

    print(dtb)
    return dtb

def quantile(dist):
    '''
    返回分布的分位数
    :param dist: 分布
    :return: 分位数
    '''
    from scipy.stats import gamma,norm

    upper_quantitle = 0

    if dist is not None:
        dist_name = list(dist.keys())[0]
        if dist_name == "gamma":
            values = dist["gamma"]
            a = values['a']
            loc = values['loc']
            scale = values['scale']
            upper_quantitle = gamma.ppf(0.9,a=a,scale=scale)

        elif dist_name == "norm":
            values = dist["norm"]
            loc = values['loc']
            scale = values['scale']
            upper_quantitle = norm.ppf(0.9,loc,scale)

    print(upper_quantitle)
    return upper_quantitle



if __name__ == "__main__":
    from scipy import stats
    data = stats.gamma.rvs(2, loc=1.5, scale=2, size=10000)
    # data = stats.norm.rvs(size=1000)
    # data = stats.poisson.rvs(mu=10,size=1000)
    dtb = dist(data,0.5)
    quantile(dtb)