Equivalent python code of "normfit" in Matlab

Question

Dose anyone know how the Confidence interval for the standard deviation is calculated in normfit function in Matlab? I need a python code to calculate such a parameter. In MATLAB, normfit returns 4 parameter, mean, std and Confidence interval of mean (muCI) and Confidence interval of standard deviation (sigmaCI).

[muHat,sigmaHat,muCI,sigmaCI] = normfit(x)

The python code below gives three parameters, muHat, sigmaHat and muCI. But I need the confidence interval of std (sigmaCI) in Python.

def function(data, confidence= 0.95):
    a = 1.0 * np.array(data)
    n = len(a)
    m ,se =  np.mean(a), scipy.stats.sem(a)
    h = se * scipy.stats.t.ppf((1 + confidence) / 2., n - 1)
    sigma = np.std(data, ddof=1)
    return m, sigma, [m - h, m + h]

Hope this be useful: https://stackoverflow.com/questions/15033511/compute-a-confidence-interval-from-sample-data — BarzanHayati, Jun 04 '19 at 08:47
https://machinelearningmastery.com/confidence-intervals-for-machine-learning/ — BarzanHayati, Jun 04 '19 at 08:48
The problem is that in most of the documents CI is calculated only for mean and not for the standard deviation which I need. The code you sent, only calculates the meanCI and not sigmaCI. — Magi, Jun 04 '19 at 08:55
You insert chi square from scipy and use this command: SQRT(((N-1)*s^2)/(CHISQ.INV((1-Confidence)/2, N-1))) — BarzanHayati, Jun 04 '19 at 09:15

Magi · Accepted Answer · 2019-06-04T09:54:25.563

0

Here is the normfit function in Python.

 def normfit(self, data, confidence=0.95):
        a = 1.0 * np.array(data)
        n = len(a)
        m, se = np.mean(a), scipy.stats.sem(a)
        h = se * scipy.stats.t.ppf((1 + confidence) / 2., n - 1)
        var = np.var(data, ddof=1)
        varCI_upper = var * (n - 1) / (scipy.stats.chi2.ppf((1-confidence) / 2, n - 1))
        varCI_lower = var * (n - 1) / (scipy.stats.chi2.ppf(1-(1-confidence) / 2, n - 1))
        sigma = np.sqrt(var)
        sigmaCI_lower = np.sqrt(varCI_lower)
        sigmaCI_upper = np.sqrt(varCI_upper)

        return m, sigma, [m - h, m + h], [sigmaCI_lower, sigmaCI_upper]

edited Jun 04 '19 at 09:54

answered Jun 04 '19 at 09:43

Magi

341
2
15

If you are sure that your code and your answer is correct, please set is as correct answer. – BarzanHayati Jun 04 '19 at 13:24
I am not able to to accept my own answer. But I am 100 % sure that code is correct. Maybe other readers can do so. – Magi Jun 05 '19 at 11:45
You need to have a minimum credit for confirm your answer as correct. Or after 24 hours. – BarzanHayati Jun 05 '19 at 17:09

Equivalent python code of "normfit" in Matlab

1 Answers1