Please enable JavaScript.

Coggle requires JavaScript to display documents.

SIMULATION ANALYTICS 1 - Coggle Diagram

- - - - has queueing/interaction
      - needs softwares
    - - no queueing/intreaction
      - can be done spreadsheets
  - - - underage, overage = cost of odering over or under the demand (under demand = opp cost)
  - - - random var -> function -> prob distribution: how comp generate data
    - - = sampling
      - take average and ignore vảiablity can be dangerous
    - - Needed to generate random input for simulation
        
        pseudo-random: randomize Uniform(0,1)
        
        good for debugging
        
        Midsquare method: square then take middle to make next random number
      - types
        
        Linear Congruential Generator
        
        Zi=(a*Zi-1 + c)(mod m)
        
        Means Zi = leftover after dividing (a*Zi-1 + c) by m
      - RANDOM VARIATE GENERATION
        
        How Works
        
        Random number generate probability,
        
        each value has different prob
        
        so from randon number R[0,1] -> value
        
        Essentially, every type of distribution works this way (the example was uniform, but could be exponential, ...)
        
        there is cheat sheet with different generation
        
        if want to use ofther function that cant be inversed, like normal distribution, can use Z value
- - - - = a/E (1 person arrive evry E mins)
  - - - variability
      - inventory (queue length)
      - utilization
  - - - average and interval
        
        use the table thing, with degree of freedom
  - - - 1 line multiple servers better
      - because "Risk pooling" -> more queue means more variability
- - - - empirical based on past, not as good as theorical
    - - basically plot data of observed against the model to see if F is the model
      - Chi-square test
        
        compare number of obersavations in an interval vs expected obs in the model distribution
        
        Wait time example in brightspace show clearly
    - - Arena has that function - Input Analyzer
  - - - random input = random output
      - needs to evaluate/compare different setup/search optimal setup
    - - vs steady state, but focus on terminating this course
        
        steady state = estimate over long run by average
      - has initial and stopping conditions
    - - how long system runs before start collecting data
    - - confidence interval half-width = h
      - Change t value to z for whatever reason
      - example slide 11 lesson 12
  - - - check w others, animation
      - put in different values (half, extreme,...)
      - That green check mark
    - - do thr out projects
      - face validity = talk to experts
      - validate assumptions
      - Sensitivity analysis
        
        do outputs change the way u want when u change input
        
        split into training and test data
    - - ?
        
        same conditions for different simulations
      - common random number to make sure dont differ too much