07-mle-and-mm

[sallymeeyan]

FAIL

• Please print/report all estimated parameters, not just the estimated medians, but all parameters - I need to know the exact values for grading.
• The QQ plot should be theoretic quantiles vs sample quantiles, that is x = theoretic quantile, and y = sample quantiles. I think you got theoretical quantiles correct in qplot, but the sample quantiles should be from the data, either gh or ht, not from Y, which is also an estimated distribution and thus theoretical.
• In MM of gamma distribution for gh, the estimated pdf and CDF are straight lines - check your results/estimated values/code. The estimated values may be off - which can be seen from the estimated median - 866.17.
• For MM of Weibull distribution for gh, could you show the code you used to solve for the equations so to get scale = 6.1516, shape = 6.3478?
• In MM of gamma distribution for ht, the estimated pdf and CDF are straight lines - check your results/estimated values/code. The estimated values may be off - which can be seen from the estimated median - 232345.11.
• Again, you need to show the code for how you get the MM of Weibull for ht.
• In MM of Weibull distribution for ht, the estimated pdf and CDF are straight lines - check your results/estimated values/code. The estimated median is off.
• Your work is well-organized with good headings. I appreciate that you have made each section very easy to locate.

[MUbarak123-56]

Hello @sallymeeyan

Thanks for the feedback. I would like to receive better clarification of your feedback. I have sighted some misunderstandings I have with your feedback below:

• First of all, which parameters would you like me to report? Is it shape and scale for weibull and gamma and mean and standard deviation for normal distribution? The report stated that we should provide the estimates for those parameters under different distributions which I did but may not have reported.

• For the method of moments estimations for both gamma and weibull distribution, I calculated them using the standard formulas that were provided to us in class, and got a distribution that did not fit with the original samples. Isn't the purpose of this deliverable to showcase the fact that there might be distinctions when using method of moments via different distributions to estimate the parameters needed to describe a distribution? Hence, I got set of values that were inconsistent with the original for those ones which I believe to be expected.

• For the calculations of my method of moments on weibull distribution for both ht and gh, you should be able to see how I used simultaneous equations to obtain both of them in the deliverable.

[sallymeeyan]

• Parameters to report: normal : mean and sd, gamma: shape and scale(or rate), weibull: shape and scale, for both MM and MLE.
• Since you did not report all parameters, I don't know the exact values of your estimates, but if your estimates are close, you should not see very bad fits as you have in the plots. And the estimated meidans should not be way off (compared to the sample median) - that's why I suspect something went wrong with your estimation.
• I can't find the code you used to solve the equations in the html file. If you can let me know where is it that will be great..
• If it's still not clear enough for you, message me and we can arrange something else to work it out.

[MUbarak123-56]

I used wolfram alpha to solve the equations in the html file. Is that fine?

[MUbarak123-56]

I realized my mistake for the charts, I was placing in shape in place of scale. The values are correct. Thanks for the feedback.

[sallymeeyan]

I used wolfram alpha to solve the equations in the html file. Is that fine?

OK, that's fine.

[sallymeeyan]

PASS

Very good!