拟合 scipy 分布并使用 nd numpy 数组检索 cdf

Question

我想拟合

gamma

分布并使用 nd numpy 数组（xarray DataArray）上的拟合参数获得

cdf

。目前，我可以用一维 numpy 数组来实现，但无法在 nd 数组上实现。非常欢迎任何其他方法。

这里是示例代码。

import xarray as xr
import numpy as np
import pandas as pd
import scipy.stats as st
# make some random data
random_data=np.random.randint(1,100,(100,50,50))
# Make ramdom datetime 
time_dim=pd.date_range("2000-01-01",periods=len(random_data),freq='MS')
# Make DataArray
test_data=xr.DataArray(random_data,dims=("time","y","x"))
# Fit gamma distribution to one data point. It works
x=test_data.isel(x=12,y=12)
shape,loc,scale=st.gamma.fit(x, scale=np.std(x))
cdf=st.gamma.cdf(x, shape,loc,scale)

test_data["time"]=time_dim
# Fitting gamma distribution and calculate cdf to each data point as time dimension. But it failed.
gamma_fit=lambda data,std_scale: st.gamma.fit(data, scale=std_scale)
std_scale=test_data.std("time")

shape, loc, scale=xr.apply_ufunc(gamma_fit,test_data, std_scale)
cdf = st.gamma.cdf(test_data, pars, loc=loc, scale=scale)