# Statistical tests

Six Sigma – iSixSigma Forums Old Forums General Statistical tests

Viewing 3 posts - 1 through 3 (of 3 total)
• Author
Posts
• #38131

Ram
Participant

Question – I have continous data for a database which shows a database size increase over a period of 3 months. I have my null and alternative hypothesis. I need to show the data is normal and independent first. Normality test (using anderson darling) is not giving the right output (i.e. does not pass the fat pencil test). Tried box-cox transformation and lamda value and conversion does not help either – so the best way would  be to do the non-parametric test – so what are the tests that i can conduct for this? Any ideas please.
2nd question – Comparitive study after implement the purge solution – how do i statistically prove that there is a difference – just draw a simple differential histogram graph? if yes, then how do we do normaility test? – do we need to compare data for each table ? we cannot just compare the size of the database since the database will again grow from a particular day when it will show the dip. What would be best approach – any advice please?
rgds
ram

0
#113736

kate
Participant

Ram
To answer your question 1. You could use Mood’s median test to conpare the two polulation .when the data is non normal
2. To prove there is a satistical difference use hypothesis test .. first check for normality . if data is normal and if there are 2 samples use 2 sample T test . if there are more than 2 samples use One way Anova .. and if the data is non normal use again Mood’s median test to compare the Median .. all this test is for means/median for varience use HOV or test of varience test ..
hope this helps

0
#113737

kate
Participant

Ram
To answer your question 1. You could use Mood’s median test to conpare the two polulation .when the data is non normal
2. To prove there is a satistical difference use hypothesis test .. first check for normality . if data is normal and if there are 2 samples use 2 sample T test . if there are more than 2 samples use One way Anova .. and if the data is non normal use again Mood’s median test to compare the Median .. all this test is for means/median for varience use HOV or test of varience test ..
hope this helps

0
Viewing 3 posts - 1 through 3 (of 3 total)

The forum ‘General’ is closed to new topics and replies.