SEP DGEST

NAME OF THE SUBJECT:

Simulation
Subject key:
Dc61
SCHEDULE:
Monday - Wednesday 09:00/11:00
NAME OF THE TEACHER:
Ing. Oscar A. Garizurieta Castillo

NAME OF THE TEAM MEMBERS:

Miriam Martínez Barajas
Zambrano Rosas Karla Madeleine
From the Cruz Mote Miguel Ángel
Reyna Rodríguez Jorge Eduardo 14500184

1
Index:

4.5.1 DESCRIPTIVE STATISTICS …………………………………………………………………………… Page 3

4.5.2 KOLMOGOROV-SMIRNOV TEST ……………………………………………………………… Page 4

4.5.3 LARGE SAMPLES: KARL-PEARSON TEST FOR FITTING A DISTRIBUTION OF

HYPOTHETICAL PROBABILITIES, DISCRETE OR CONTINUOUS (IN SPREADSHEET OR WITH PACKAGE)
STATISTICAL) ……………………………………………………………………………………………………. Page 6

4.5.4 OTHER TESTS: ANDERSON-DARLING, G TEST, FOR EXAMPLE. ……………………… Page 6

4.6 SIMULATION OF THE RANDOM BEHAVIORS OF THE PROJECT Y SU

VERIFICATION ……………………………………………………………………………………………………. Page 7

2
4.5.1 DESCRIPTIVE STATISTICS

Procedures that serve to present quantitative data in a summarized way revealing the
characteristics (mean and standard deviation for example) of a data distribution.

The information provided by descriptive statistics can often be easily and effectively conveyed.
through a variety of graphic methods such as: trend graphs (tracing a characteristic of
interest over a period); scatter plots (relationship between two variables, one on the Y axis and the other
about the X axis); the histograms (distribution of the values of a characteristic).

Other graphic methods that aid in data interpretation and analysis include: bar charts,
graphs with special scales (such as probability ones) and graphs involving dimensions and variables
multiple.

Graphical methods are useful for summarizing and presenting complex data or relationships between data.
effectively, particularly for a non-specialized audience.

Some uses

•Summarize and characterize data. It is usually the initial step in the quantitative data analysis and
It often constitutes the first step towards the use of other statistical procedures.

The characteristics of the sample data can serve as a basis for making inferences regarding the
characteristics of the populations from which the sample was taken.

Some benefits

Relatively simple and efficient way to summarize and characterize data.

It offers a convenient way to present information. In particular, graphical methods are a

a very effective way to present data, and to communicate information.

Descriptive statistics is potentially applicable to all situations that involve the use of data.
It can assist in the analysis and interpretation of data and is a valuable aid in decision-making.

Limitations and precautions

Descriptive statistics provide quantitative measurements of characteristics (such as the

average and the standard deviation) of sample data. However, these measurements are subject to the
limitations of the sample size and the sampling method used.

Application examples

describe the behavior of a process parameter, such as the temperature of a chamber;

characterize the delivery time or the response time in a service;

summarize customer survey data, such as customer satisfaction;

•visualize the distribution of a characteristic of a process using a histogram, against the limits of
specification for that feature;

visualize the performance result of a product over a period using a trend chart.

3
4.5.2 KOLMOGOROV-SMIRNOV TEST

The Kolmogorov-Smirnov goodness of fit test is an alternative to test that a sample

it "comes" from a continuous distribution (normal). This test is based on the comparison between the function
cumulative distribution of a theoretical distribution with the cumulative distribution function of the sample.

If the theoretical and sample cumulative distribution functions are not significantly different, then
we say that the sample comes from the distribution whose cumulative distribution function is Ft(x). However,
if the differences between the cumulative distribution functions are too large for them not to be
due solely to chance, we reject Ho

The steps to follow in the Kolmogorov-Smirnov goodness of fit test are as follows:

Plantear la hipótesis: Ho: Fm(X)=Ft(X) para todo X E R; Ha: Fm(X)=Ft(X), por lo menos para un X.
Calculate all the Fm(X) values of the sample X1,X2…,Xn.
Determine the maximum deviation, which is given by the supremum of the absolute values of the
differences between the values of the theoretical cumulative function and those of the sample.
Choose a level of significance
According to the result, the decision is made
The assumptions in the Kolmogorov-Smirnov goodness of fit test are:
Random Samples
2. The population should be continuous in the observed variable.
3. The test is not valid if one or more parameters have to be estimated using the data from the
sample.

4
KOLMOGOROV-SMIRNOV TABLE

5
4.5.3 LARGE SAMPLES: KARL-PEARSON TEST FOR GOODNESS OF FIT
FROM A HYPOTHETICAL PROBABILITY DISTRIBUTION, DISCRETE OR
CONTINUE (IN SPREADSHEET OR WITH STATISTICAL PACKAGE)
This is contained in a record from 1900 and is introduced to provide a measure of the fit between a distribution.
of probability and a sample. The idea is, given the sample (x1, ..., xn) and the distribution f(x|θ) to construct the
statistic

What is distributed χ² k−1, if the sample comes from the distribution. Where a partition is assumed to have been made.
of k elements in the distribution path, with which the values yi, the observed frequencies of the
xi in the element i of the partition can be assumed to have a multinomial distribution, and yi are the frequencies.
observed under the hypothesis that the sample distribution is f(x|θ). The procedure would be generalized
to the problems of homogeneity and to the contingency tables, by K. Pearson himself and by his
disciples, Edgeworth and Yule, culminating in the later works of Fisher. Relevant information from
This evolution can be seen in Stigler (1986), the development of methods can be seen in Gómez Villegas.
(2005).

4.5.4 OTHER TESTS: ANDERSON-DARLING, G TEST, FOR EXAMPLE.

The Anderson-Darling statistic measures how well the data follow a specific distribution. For a
data set and distribution in particular, the better the distribution fits the data, the lower it will be
this statistic. For example, you can use the Anderson-Darling statistic to determine if the
data satisfy the normality assumption for a t-test.

The hypotheses for the Anderson-Darling test are:

H0The data follow a specified distribution

H1The data do not follow a specified distribution

Use the corresponding p-value (if available) to test if the data comes from the distribution.
selected. If the p-value is less than a chosen significance level (usually 0.05 or 0.10), then
reject the null hypothesis that the data comes from that distribution. Minitab does not always show a
p-value for the Anderson-Darling test, because it does not exist mathematically for certain cases.

You can also use the Anderson-Darling statistic to compare the fit of several distributions with
the purpose of determining which is the best. However, to conclude that a distribution is the best, the statistician
the Anderson-Darling should be substantially lower than the others. When the statistics are close
between them, additional criteria, such as probability graphs, should be used to choose between them.

6
 Example
of

comparison of distributions

These probability graphs are for the same data. Both the normal distribution and the distribution of
3-parameter Weibull provides a good fit to the data.

Minitab calculates the Anderson-Darling statistic using the weighted squared distance from the line
adjusted from the probability graph (based on the chosen distribution and using the estimation method
of maximum likelihood or the least squares estimates) and the non-parametric step function.
calculus has a greater weight in the tails of the distribution.

4.6 SIMULATION OF RANDOM BEHAVIORS OF

PROJECT AND ITS VERIFICATION.

The figure shows the fundamental phases that constitute a simulation study:

7
Simulation is an increasingly used technique in the study of complex systems. Among the
Arguments in favor of the use of simulation include the following:

Most real complex systems with stochastic elements cannot be described with
sufficient accuracy through a mathematical model that can be solved analytically. Therefore, with
Frequency, simulation is the only possible method of studying such systems.

Simulation allows estimating the behavior of an existing system under a planned set of
operational conditions.

Through simulation, alternative designs (or alternative operating policies) can be compared.
a specific design) to specify which one meets the objectives most adequately
formulated.

In the simulation, one can have much better control over the conditions of the experiment than if one
to be realized about the system itself.

- Simulation allows studying a system whose evolution is very lengthy in time (for example, a
economic system) in a short period of time. Alternatively, it also allows for studying in a way
detailed the evolution of a system in a short period of time.

Experience shows that there are a number of mistakes that are often made when carrying out a
simulation study. Among them, the following can be highlighted: Discrete system simulation - Not defined
correctly the objectives of the study.

8
Set an inadequate level of detail in the model.

Treat the simulation study as if it were primarily a complicated programming exercise.

Using a business simulation software that cannot adequately reflect the logic of
functioning of the model.

− Inappropriately using animation.

− Inadequately determining the sources of randomness in the real system.

Use probability distribution functions different from those corresponding to real phenomena.
that they want to simulate.

Analyze the resulting data from the simulation considering, in the statistical formulas used, that
all values are independent.

Perform a smaller number of repetitions than necessary and consider the results obtained to be significant.