Challenge #141: Examination Data Simulation

[tristank]

I have never used this tool before so it took some playing around and I don't know if it was the best approach but I got a solid solution. Cool stuff!

Spoiler

[aatalai]

used motne carlo so wasn't too bad

[lars3033]

I used the Central Limit Theorem (CLT), the uniform random number function RAND(), an iterative macro, and some trial and error.

The CLT proposes that the mean of a sample of data from any distribution is itself distributed normally regardless of the distribution of the underlying data.

In other words, if you take a sample, say 1,000 values, of any data population, then take the mean of that sample, and repeat this many times, then the distribution of those means will be normal.

1)  Randomly generate 1,000 numbers with low and high values that will result in an average (over those 1,000 numbers) between 50 and 90.  This was the hard part.

In a formula tool  >>>  Average(70 - 850 * RAND(), 70 + 850 * RAND())

2)  Iterate 1,000 of these 1,000 observation samples and take the mean from each, giving you 1,000 means.

3)  You're done!  Just build a histogram of the means ...

[JoshuaM]

Simulation Achieved !

Solution:

Spoiler

Explanation:

Spoiler

The Simulation tool was configured using a Monte Carlo Sampling method. The chunk size and seed were left at their default values. Yet the number of iterations was changed to 1000 to represent the number of students in the data set. Sample parametrically was selected as no data was entered into the input nodes. I joined the data set via position as the number of records aligned. 
Lastly I aggregated the sample test scores and counted the number of students. The data was visualized via the interactive chart tool and given a title for each component.