Ufdjrw

Why is there an extra space when I type "ls" on the Desktop?

Is "cogitate" an appropriate word for this?

Is every open circuit a capacitor?

Why can't we use freedom of speech and expression to incite people to rebel against government in India?

Why would the IRS ask for birth certificates or even audit a small tax return?

I can't die. Who am I?

How do you make a gun that shoots melee weapons and/or swords?

What does "rhumatis" mean?

Does the in-code argument passing conventions used on PDP-11's have a name?

Did Amazon pay $0 in taxes last year?

Can a Mexican citizen living in US under DACA drive to Canada?

Quitting employee has privileged access to critical information

Are Wave equations equivalent to Maxwell equations in free space?

Can a space-faring robot still function over a billion years?

Short story about an infectious indestructible metal bar?

Why do phishing e-mails use faked e-mail addresses instead of the real one?

Remove object from array based on array of some property of that object

What is the meaning of option 'by' in TikZ Intersections

New invention compresses matter to produce energy? or other items? (Short Story)

Problems with rounding giving too many digits

Called into a meeting and told we are being made redundant (laid off) and "not to share outside". Can I tell my partner?

3.5% Interest Student Loan or use all of my savings on Tuition?

Where do you go through passport control when transiting through another Schengen airport on your way out of the Schengen area?

Error in TransformedField

Problems performing a lognormal fitting with Scipy

1

I have been reading a lot of past question regarding this topic, however I have not managed to find a solution to this problem. When using python scipy.stats.lognormal.fit for a set of data (Y), the output parameters are the shape, loc and scale respectively. The shape it is supposed to be equal to the standard deviation of the normally distributed random variable X such that exp(X)= Y, and the scale equal to the exp(mu), were mu is the mean of the same normally distributed random variable X. I have three sets of data for which I am performing the fitting, however there is particularly one of them giving me trouble, and I have not been able to find the reason. I cannot share the code as the sample is very specific and has a size of 270, so I am reading it directly from a file, but the problem is that I am getting different values of standard deviation= scale and mu= ln(scale) by performing the fitting in two different ways: One fitting directly the data Y by using lognorm.fit, and another one performing the fitting into the normally distributed sample X. I checked with another softwares (Matlab and Excel), and the second option is giving the right results, so my guess is that it is related with the convergence or a particular characteristic of the lognorm.fit, however I have tried several things unsuccesfully, as for example using an optimzer:

#def opti_wrap(fun, x0, args, disp=0, **kwargs):
# return minimize(fun, x0, args=args, method='SLSQP', 
# tol=1e-12, options='maxiter': 1000).x

or playing with the loc value when doing the lognorm.fit.

Any ideas what could be the issue?

python scipy

share|improve this question

asked yesterday

derekrpderekrp

61

New contributor

derekrp is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  Managed to fix the problem. Apparently the loc parameter is the problem. Depending of the value it gives problems with the fitting process. In order to fix the problem, I fixed it using floc= 0, and now the results are matching the fitting process by using the normalized set of data.
  
  – derekrp
  yesterday
  
  

  
  
  
  

add a comment | 

1

I have been reading a lot of past question regarding this topic, however I have not managed to find a solution to this problem. When using python scipy.stats.lognormal.fit for a set of data (Y), the output parameters are the shape, loc and scale respectively. The shape it is supposed to be equal to the standard deviation of the normally distributed random variable X such that exp(X)= Y, and the scale equal to the exp(mu), were mu is the mean of the same normally distributed random variable X. I have three sets of data for which I am performing the fitting, however there is particularly one of them giving me trouble, and I have not been able to find the reason. I cannot share the code as the sample is very specific and has a size of 270, so I am reading it directly from a file, but the problem is that I am getting different values of standard deviation= scale and mu= ln(scale) by performing the fitting in two different ways: One fitting directly the data Y by using lognorm.fit, and another one performing the fitting into the normally distributed sample X. I checked with another softwares (Matlab and Excel), and the second option is giving the right results, so my guess is that it is related with the convergence or a particular characteristic of the lognorm.fit, however I have tried several things unsuccesfully, as for example using an optimzer:

#def opti_wrap(fun, x0, args, disp=0, **kwargs):
# return minimize(fun, x0, args=args, method='SLSQP', 
# tol=1e-12, options='maxiter': 1000).x

or playing with the loc value when doing the lognorm.fit.

Any ideas what could be the issue?

python scipy

share|improve this question

asked yesterday

derekrpderekrp

61

New contributor

derekrp is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

• 
  
  
  

  
  

  
  
  
  
  
  

  
  
  Managed to fix the problem. Apparently the loc parameter is the problem. Depending of the value it gives problems with the fitting process. In order to fix the problem, I fixed it using floc= 0, and now the results are matching the fitting process by using the normalized set of data.
  
  – derekrp
  yesterday
  
  

  
  
  
  

add a comment | 

I have been reading a lot of past question regarding this topic, however I have not managed to find a solution to this problem. When using python scipy.stats.lognormal.fit for a set of data (Y), the output parameters are the shape, loc and scale respectively. The shape it is supposed to be equal to the standard deviation of the normally distributed random variable X such that exp(X)= Y, and the scale equal to the exp(mu), were mu is the mean of the same normally distributed random variable X. I have three sets of data for which I am performing the fitting, however there is particularly one of them giving me trouble, and I have not been able to find the reason. I cannot share the code as the sample is very specific and has a size of 270, so I am reading it directly from a file, but the problem is that I am getting different values of standard deviation= scale and mu= ln(scale) by performing the fitting in two different ways: One fitting directly the data Y by using lognorm.fit, and another one performing the fitting into the normally distributed sample X. I checked with another softwares (Matlab and Excel), and the second option is giving the right results, so my guess is that it is related with the convergence or a particular characteristic of the lognorm.fit, however I have tried several things unsuccesfully, as for example using an optimzer:

#def opti_wrap(fun, x0, args, disp=0, **kwargs):
# return minimize(fun, x0, args=args, method='SLSQP', 
# tol=1e-12, options='maxiter': 1000).x

or playing with the loc value when doing the lognorm.fit.

Any ideas what could be the issue?

python scipy

share|improve this question

asked yesterday

derekrpderekrp

61

New contributor

derekrp is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

#def opti_wrap(fun, x0, args, disp=0, **kwargs):
# return minimize(fun, x0, args=args, method='SLSQP', 
# tol=1e-12, options='maxiter': 1000).x

share|improve this question

asked yesterday

derekrpderekrp

61

New contributor

derekrp is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

share|improve this question

asked yesterday

derekrpderekrp

61

New contributor

derekrp is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

asked yesterday

derekrpderekrp

61

New contributor

derekrp is a new contributor to this site. Take care in asking for clarification, commenting, and answering.
Check out our Code of Conduct.

asked yesterday

derekrpderekrp

61