why does z-test indicate significantly different for 2 distributions that looks similar (hypothesis testing)












0















I have two distributions below (Kaggle dataset: Rossman sales) that look similar visually: Sales on normal days & sales on school holiday.



enter image description here



However, they seems to fail z-test (hypothesis testing) in Python - why is that so?



How should I perform the statistical test (z-test) in Python? Should I use pooled or unequalvar (should I use same variance or different)? I also found out that switching School_hol_sales and Normal_day_sales in the code below yield different results and I am not sure why.



School_hol_sales = df[(df.Open==1)&(df.SchoolHoliday==1)&(df.StateHoliday=='0')&(df.Promo==0)].Sales
Normal_day_sales = df[(df.Open==1)&(df.SchoolHoliday==0)&(df.StateHoliday=='0')&(df.Promo==0)].Sales
School_hol_sales.mean(), Normal_day_sales.mean() # (6230.4, 5904.6)
School_hol_sales.std(), Normal_day_sales.std() # (2841.8, 2602.9)


# which is the correct one?
import statsmodels.stats.api as sms

cm = sms.CompareMeans(sms.DescrStatsW(School_hol_sales), sms.DescrStatsW(Normal_day_sales))
z, pval = cm.ztest_ind(alternative='larger', usevar='unequal')
print('z: {} , pval: {}'.format(z, pval))


from statsmodels.stats.weightstats import ztest

z, pval = ztest(School_hol_sales,Normal_day_sales, alternative='larger', usevar='pooled', ddof=1.0)
print('z: {} , pval: {}'.format(z, pval))


Output:



z: 28.53350149055591 , pval: 2.2504631945823565e-179
z: 30.17089944207645 , pval: 2.853425122518376e-200









share|improve this question























  • If I may ask you, what exactly do you want to test? Is it the possibility that the two samples you mention are actually samples of the same population and thus have the same distribution?

    – DavidPM
    Nov 28 '18 at 16:34






  • 1





    Statistical significance is not the same thing as practical significance. If two means are actually different, in any degree no matter how small, a significance test will almost certainly fail given sufficiently large samples; this is a well-known feature or bug, depending on one's point of view. My advice is dump all significance tests and work only with practical significance, i.e. assess a value (in money, time, resources, whatever) for your actions and go from there.

    – Robert Dodier
    Nov 28 '18 at 19:28


















0















I have two distributions below (Kaggle dataset: Rossman sales) that look similar visually: Sales on normal days & sales on school holiday.



enter image description here



However, they seems to fail z-test (hypothesis testing) in Python - why is that so?



How should I perform the statistical test (z-test) in Python? Should I use pooled or unequalvar (should I use same variance or different)? I also found out that switching School_hol_sales and Normal_day_sales in the code below yield different results and I am not sure why.



School_hol_sales = df[(df.Open==1)&(df.SchoolHoliday==1)&(df.StateHoliday=='0')&(df.Promo==0)].Sales
Normal_day_sales = df[(df.Open==1)&(df.SchoolHoliday==0)&(df.StateHoliday=='0')&(df.Promo==0)].Sales
School_hol_sales.mean(), Normal_day_sales.mean() # (6230.4, 5904.6)
School_hol_sales.std(), Normal_day_sales.std() # (2841.8, 2602.9)


# which is the correct one?
import statsmodels.stats.api as sms

cm = sms.CompareMeans(sms.DescrStatsW(School_hol_sales), sms.DescrStatsW(Normal_day_sales))
z, pval = cm.ztest_ind(alternative='larger', usevar='unequal')
print('z: {} , pval: {}'.format(z, pval))


from statsmodels.stats.weightstats import ztest

z, pval = ztest(School_hol_sales,Normal_day_sales, alternative='larger', usevar='pooled', ddof=1.0)
print('z: {} , pval: {}'.format(z, pval))


Output:



z: 28.53350149055591 , pval: 2.2504631945823565e-179
z: 30.17089944207645 , pval: 2.853425122518376e-200









share|improve this question























  • If I may ask you, what exactly do you want to test? Is it the possibility that the two samples you mention are actually samples of the same population and thus have the same distribution?

    – DavidPM
    Nov 28 '18 at 16:34






  • 1





    Statistical significance is not the same thing as practical significance. If two means are actually different, in any degree no matter how small, a significance test will almost certainly fail given sufficiently large samples; this is a well-known feature or bug, depending on one's point of view. My advice is dump all significance tests and work only with practical significance, i.e. assess a value (in money, time, resources, whatever) for your actions and go from there.

    – Robert Dodier
    Nov 28 '18 at 19:28
















0












0








0








I have two distributions below (Kaggle dataset: Rossman sales) that look similar visually: Sales on normal days & sales on school holiday.



enter image description here



However, they seems to fail z-test (hypothesis testing) in Python - why is that so?



How should I perform the statistical test (z-test) in Python? Should I use pooled or unequalvar (should I use same variance or different)? I also found out that switching School_hol_sales and Normal_day_sales in the code below yield different results and I am not sure why.



School_hol_sales = df[(df.Open==1)&(df.SchoolHoliday==1)&(df.StateHoliday=='0')&(df.Promo==0)].Sales
Normal_day_sales = df[(df.Open==1)&(df.SchoolHoliday==0)&(df.StateHoliday=='0')&(df.Promo==0)].Sales
School_hol_sales.mean(), Normal_day_sales.mean() # (6230.4, 5904.6)
School_hol_sales.std(), Normal_day_sales.std() # (2841.8, 2602.9)


# which is the correct one?
import statsmodels.stats.api as sms

cm = sms.CompareMeans(sms.DescrStatsW(School_hol_sales), sms.DescrStatsW(Normal_day_sales))
z, pval = cm.ztest_ind(alternative='larger', usevar='unequal')
print('z: {} , pval: {}'.format(z, pval))


from statsmodels.stats.weightstats import ztest

z, pval = ztest(School_hol_sales,Normal_day_sales, alternative='larger', usevar='pooled', ddof=1.0)
print('z: {} , pval: {}'.format(z, pval))


Output:



z: 28.53350149055591 , pval: 2.2504631945823565e-179
z: 30.17089944207645 , pval: 2.853425122518376e-200









share|improve this question














I have two distributions below (Kaggle dataset: Rossman sales) that look similar visually: Sales on normal days & sales on school holiday.



enter image description here



However, they seems to fail z-test (hypothesis testing) in Python - why is that so?



How should I perform the statistical test (z-test) in Python? Should I use pooled or unequalvar (should I use same variance or different)? I also found out that switching School_hol_sales and Normal_day_sales in the code below yield different results and I am not sure why.



School_hol_sales = df[(df.Open==1)&(df.SchoolHoliday==1)&(df.StateHoliday=='0')&(df.Promo==0)].Sales
Normal_day_sales = df[(df.Open==1)&(df.SchoolHoliday==0)&(df.StateHoliday=='0')&(df.Promo==0)].Sales
School_hol_sales.mean(), Normal_day_sales.mean() # (6230.4, 5904.6)
School_hol_sales.std(), Normal_day_sales.std() # (2841.8, 2602.9)


# which is the correct one?
import statsmodels.stats.api as sms

cm = sms.CompareMeans(sms.DescrStatsW(School_hol_sales), sms.DescrStatsW(Normal_day_sales))
z, pval = cm.ztest_ind(alternative='larger', usevar='unequal')
print('z: {} , pval: {}'.format(z, pval))


from statsmodels.stats.weightstats import ztest

z, pval = ztest(School_hol_sales,Normal_day_sales, alternative='larger', usevar='pooled', ddof=1.0)
print('z: {} , pval: {}'.format(z, pval))


Output:



z: 28.53350149055591 , pval: 2.2504631945823565e-179
z: 30.17089944207645 , pval: 2.853425122518376e-200






python statistics data-science hypothesis-test






share|improve this question













share|improve this question











share|improve this question




share|improve this question










asked Nov 28 '18 at 10:29









jasonlcy91jasonlcy91

779




779













  • If I may ask you, what exactly do you want to test? Is it the possibility that the two samples you mention are actually samples of the same population and thus have the same distribution?

    – DavidPM
    Nov 28 '18 at 16:34






  • 1





    Statistical significance is not the same thing as practical significance. If two means are actually different, in any degree no matter how small, a significance test will almost certainly fail given sufficiently large samples; this is a well-known feature or bug, depending on one's point of view. My advice is dump all significance tests and work only with practical significance, i.e. assess a value (in money, time, resources, whatever) for your actions and go from there.

    – Robert Dodier
    Nov 28 '18 at 19:28





















  • If I may ask you, what exactly do you want to test? Is it the possibility that the two samples you mention are actually samples of the same population and thus have the same distribution?

    – DavidPM
    Nov 28 '18 at 16:34






  • 1





    Statistical significance is not the same thing as practical significance. If two means are actually different, in any degree no matter how small, a significance test will almost certainly fail given sufficiently large samples; this is a well-known feature or bug, depending on one's point of view. My advice is dump all significance tests and work only with practical significance, i.e. assess a value (in money, time, resources, whatever) for your actions and go from there.

    – Robert Dodier
    Nov 28 '18 at 19:28



















If I may ask you, what exactly do you want to test? Is it the possibility that the two samples you mention are actually samples of the same population and thus have the same distribution?

– DavidPM
Nov 28 '18 at 16:34





If I may ask you, what exactly do you want to test? Is it the possibility that the two samples you mention are actually samples of the same population and thus have the same distribution?

– DavidPM
Nov 28 '18 at 16:34




1




1





Statistical significance is not the same thing as practical significance. If two means are actually different, in any degree no matter how small, a significance test will almost certainly fail given sufficiently large samples; this is a well-known feature or bug, depending on one's point of view. My advice is dump all significance tests and work only with practical significance, i.e. assess a value (in money, time, resources, whatever) for your actions and go from there.

– Robert Dodier
Nov 28 '18 at 19:28







Statistical significance is not the same thing as practical significance. If two means are actually different, in any degree no matter how small, a significance test will almost certainly fail given sufficiently large samples; this is a well-known feature or bug, depending on one's point of view. My advice is dump all significance tests and work only with practical significance, i.e. assess a value (in money, time, resources, whatever) for your actions and go from there.

– Robert Dodier
Nov 28 '18 at 19:28














0






active

oldest

votes











Your Answer






StackExchange.ifUsing("editor", function () {
StackExchange.using("externalEditor", function () {
StackExchange.using("snippets", function () {
StackExchange.snippets.init();
});
});
}, "code-snippets");

StackExchange.ready(function() {
var channelOptions = {
tags: "".split(" "),
id: "1"
};
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function() {
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled) {
StackExchange.using("snippets", function() {
createEditor();
});
}
else {
createEditor();
}
});

function createEditor() {
StackExchange.prepareEditor({
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: true,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: 10,
bindNavPrevention: true,
postfix: "",
imageUploader: {
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
},
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
});


}
});














draft saved

draft discarded


















StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53517313%2fwhy-does-z-test-indicate-significantly-different-for-2-distributions-that-looks%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown

























0






active

oldest

votes








0






active

oldest

votes









active

oldest

votes






active

oldest

votes
















draft saved

draft discarded




















































Thanks for contributing an answer to Stack Overflow!


  • Please be sure to answer the question. Provide details and share your research!

But avoid



  • Asking for help, clarification, or responding to other answers.

  • Making statements based on opinion; back them up with references or personal experience.


To learn more, see our tips on writing great answers.




draft saved


draft discarded














StackExchange.ready(
function () {
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fstackoverflow.com%2fquestions%2f53517313%2fwhy-does-z-test-indicate-significantly-different-for-2-distributions-that-looks%23new-answer', 'question_page');
}
);

Post as a guest















Required, but never shown





















































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown

































Required, but never shown














Required, but never shown












Required, but never shown







Required, but never shown







Popular posts from this blog

A CLEAN and SIMPLE way to add appendices to Table of Contents and bookmarks

Calculate evaluation metrics using cross_val_predict sklearn

Insert data from modal to MySQL (multiple modal on website)