Hello Internet, I´m really confused about one thing I have been trying to solve. It is a real world situation, but I am gonna tell you just the statistical part and simplify the background and I hope you can tell me if it is right how I see it.
Let’s assume a company manufactures a product and they count the amount of products they manufacture by month. They just started one year ago (12 months and here is their record:
Now, let’s say that a percentage of their products presents a malfunction they detect after manufacturing, let’s assume they cannot stop it or prevent it and it is not related to the months. Here is the data of the defectives products:
Now the problem is: If the company knew how many products they would produce in the upcoming year each month, how could they simulate the situation the would have with the defectives?
The forecast production for the next year:
This is how I see it:
I assumed it was a normal distribution and I found the mean (26,14%) and the standard deviation (19,18%) of the percentages of defectives.
Then, performing a pseudo-Monte Carlo simulation, I used to excel to generate a random number between 0 and 1, then I used an inverser normal function that gives the value of a normal function with a probability (the random number), a mean, and a deviation input. After that, I get percentages of defective products, so I multiply them with each production month with a diferent value.
i.e:
Note that the % of defectives is generated with a simulation using the inverse of the normal distribution and due to that it generates negative numbers at some points, which of course it doesn´t make sense and I could take them just as a 0 porcentages of defectives.
What do you think? I get confused by the fact of using both percentages and probabilities and another thing which I am not very sure about is considering normal distribution, I don’t know how I could resolve it with another type of distribution.
Let’s assume a company manufactures a product and they count the amount of products they manufacture by month. They just started one year ago (12 months and here is their record:
Now, let’s say that a percentage of their products presents a malfunction they detect after manufacturing, let’s assume they cannot stop it or prevent it and it is not related to the months. Here is the data of the defectives products:
Now the problem is: If the company knew how many products they would produce in the upcoming year each month, how could they simulate the situation the would have with the defectives?
The forecast production for the next year:
This is how I see it:
I assumed it was a normal distribution and I found the mean (26,14%) and the standard deviation (19,18%) of the percentages of defectives.
Then, performing a pseudo-Monte Carlo simulation, I used to excel to generate a random number between 0 and 1, then I used an inverser normal function that gives the value of a normal function with a probability (the random number), a mean, and a deviation input. After that, I get percentages of defective products, so I multiply them with each production month with a diferent value.
i.e:
Note that the % of defectives is generated with a simulation using the inverse of the normal distribution and due to that it generates negative numbers at some points, which of course it doesn´t make sense and I could take them just as a 0 porcentages of defectives.
What do you think? I get confused by the fact of using both percentages and probabilities and another thing which I am not very sure about is considering normal distribution, I don’t know how I could resolve it with another type of distribution.
