# analysing the fiability of market shares study

#### Mineral

##### New Member
Hello everybody!

I'm a beginner data analyst in the retail/CRM business.
I have been asked about classifying the fiability of a study.

The study produces the following report about the sales percentage of each company in the retail business. Each row belongs to a company

week 1 week2 week3 week4 week5
6,6% 6,4% 7,0% 6,4% 6,5%
13,0% 13,8% 13,2% 13,1% 12,8%
2,3% 2,0% 2,1% 2,2% 2,3%
9,9% 9,5% 9,7% 9,7% 9,7%
8,7% 9,2% 9,3% 9,1% 9,5%
6,9% 6,5% 6,8% 7,2% 7,1%
11,0% 11,6% 11,2% 11,6% 11,6%
14,8% 14,2% 15,0% 15,1% 15,6%
26,7% 26,7% 25,8% 25,6% 24,9%

I know the real values of sales of two companies presented in this report.
The study is produced using a sample of 3000 families and their shopping data. I've already seen some inconsistent values applying the percentages to the absolute sales values of each of our company which is natural but I wanted to know whether these discrepancies question the study itself or if they belong to the interval assuming a normal distribution with 95% of confidence! How can I test this?

I apologize if I wasnt clear about my question!
Thank you!

#### Dr.Appalayya

##### New Member
Use real values. Find out if the values have normal distribution. This can be done with KS test or making bar chart on SPSS.

#### Mineral

##### New Member
Not a normal distribution I think

Summarizing real values data for the first week for the two companies I know real data from.

Min. :-449.00
1st Qu.: 14.83
Median : 31.98
Mean : 47.30
3rd Qu.: 62.77
Max. :4504.70

Its not a normal distribution I believe...

#### Dr.Appalayya

##### New Member
It does not look normal. Use SPSS to find out KS test statistic or Shapiro's W to find out if the data is normally distributed or not. From what you have supplied, it is not possible to give suggestions.

#### Mineral

##### New Member
kolmogorov-smirnov test for normality

One-sample Kolmogorov-Smirnov test

data: semana[, 2]
D = 0.1784, p-value < 2.2e-16
alternative hypothesis: two-sided

I've got this warning:

In ks.test(semana[, 2], "pnorm") :
cannot compute correct p-values with ties

Thanks