Predictive date using Standard Deviation

[P_Kumar]

Hi,

I have a requirement to calculate the predictive date (in C2 column of the excel attached) based on the historical receipt date of the product with 95% probability using  Standard deviation.  I have attached the sample data and in the excel and has derived the Arithmetic Stdev as 19.  I have used the Stdev function in Qlik Sense and got the same value as 19.

Can someone please advise how could I get the predictive date of 95% probability using Stdev in Qlik Sense.

Thanks- Kumar

[lorenzoconforti]

Assuming the data is normally distributed, you just need to multiply the mean of your values by +- 1.96 times the standard deviation divided by the square root of the number of observations. 

See excel attached

[P_Kumar]

Thanks a lot @lorenzoconforti  - Can you help me with following:

- When you say ' Assuming the data is normally distributed' - isn't  my data set normally distributed? if not, how should it be?

- What is the logic behind multiplying with 1.96?

I know these must be basics of Stats but would be helpful for my scenario if you could let me know. Thanks

Regards-Kumar

[lorenzoconforti]

It really depends on what you are trying to achieve. First of all you've provided very few data points so it's hard to tell what kind of distribution there is so you should think about our data and evaluate what results you usually get. 

A basic prediction is the mean itself; then, as you mentioned, you can include confidence intervals that would give you a measure of upper and lower limit within a certain probability percentage. In your case, given the data provided, we can say that on average 95% of the data points fall between 123.2 and 153.8 days

You can then further improve the model by calculating a regression function to see if there is any trend on the data

It really depends on what you are looking to achieve