on which the validity and adequacy of the model are checked. The created model must be adequate to the real economic process. If the quality of the model is unsatisfactory, they return to the second stage of modeling.
7. Stage of interpretation of simulation results.
Rice. 2.1. The main stages of econometric modeling
forecast of economic indicators characterizing the process under study (phenomenon, object);
modeling the behavior of a process (phenomenon, object) when different meanings factorial variables;
formation of management decisions.
The number of variables included in the econometric model should not be too large and should be theoretically justified. The model should lack a functional or close correlation between factor variables, which can lead to the phenomenon multicollinearity.
When forming the initial information for the econometric model, an extremely important problem is the choice of indicators that are adequate to the essence of the phenomena under study. Often, an econometric model is built precisely to express the pattern that exists between phenomena. Attention should be paid to a certain substitution of concepts, which usually occurs at the first stage of building a model during the transition from a meaningful analysis of phenomena to the formation of quantitative characteristics (indicators) reflecting their levels. In the course of meaningful analysis, the phenomenon is often considered on quality level. However, when building a model, initial information, sets of indicators are used that express these phenomena, their properties, trends in the form quantitative characteristics.
For traditional areas of research, the problem of substantiating the composition of indicators is usually considered solved. For example, in studies of labor productivity, macroeconomic analysis usually considers already established sets of
indicators, the values of which are published in statistical collections, scientific reports, etc. Their example is the output per worker as an indicator expressing the phenomenon of "labor productivity", the volume of GDP (an indicator of the efficiency of the economy), the volume of fixed assets (an indicator of the level of material security of the production process, the economy), etc. At the same time, in a number of areas of econometric research, such systems of indicators cannot be formed so unambiguously. Often the same phenomenon can be expressed by alternative variants of indicators. In the absence of objective data in econometric studies, it is allowed to replace one indicator with another that indirectly reflects the same phenomenon. For example, the average per capita income as an indicator of the material standard of living can be replaced by the average annual turnover per one inhabitant of the region, etc. The wrong choice of the indicator that represents the phenomenon under consideration in the model can significantly affect its quality, in connection with which the problem of substantiating the composition of indicators (variables) of the econometric model in practice should be treated with utmost attention.
Considering the problem of choosing a specific function type , it should be noted that in the practice of econometric research, a fairly wide range of functional dependencies between variables is used, the most often used are: linear, right semilogarithmic, sedate, hyperbolic, logarithmic hyperbolic, inverse linear (Tornquist function), function with permanent elasticity of replacement, exponential function... In practice, combinations of the above dependencies can also be encountered, for example,
Most of the functions 
with the help of a certain set of transformations, they can be reduced to a linear form. For example, if 
and 
addicted 
(7), then introducing the variables 
, we obtain expression (4) up to the transformation of the initial factors.
In practical research, often using the transformation 
and 
, the power-law model (6) is transformed to a linear form connecting the logarithms of the variables 
and 
.
However, it should be noted that in in this case, from the point of view of mathematics, such a transformation is not entirely correct due to the additivity of the error in expression (6), therefore, the values of the coefficients of the linear (with respect to the logarithms of variables) model cannot be general case to be considered equal to the corresponding values of the power analogue.
Using the example of a linear econometric model, one can imagine another form of models of this type - models in which there is no free coefficient 
:
In many practical studies, strict theoretical concepts, preliminary assumptions about the meaningful aspects of interaction between phenomena recede into the background. For them, the main thing is to build an equation that quite accurately expresses the relationships that are adequate to the trends in the change of variables. 
and 
on the time interval (1, T)... Moreover, it is often the successful form of the equation of the econometric model that forms the basis of the developed theoretical concept, which then finds its application in subsequent analysis. Obviously, the most "suitable" form provides the best approximation of the theoretical (calculated) frequencies of the values
to actual values 
.
Usually, the choice of the form of dependence is carried out on the basis of graphical analysis development trends of the corresponding processes. For example, if the variable 
and variable
changed over time according to the graphs shown in Fig. 2.2, then it is logical to assume that the dependence is hyperbolic 
.
For the graphs shown in Fig. 2.3, a logarithmic dependence is characteristic 
.
Rice. 2.2. Hyperbolic dependence
Rice. 2.3. Logarithmic dependence
Optimal composition of factors
included in the econometric model, one of the main conditions for its good quality, understood as the conformity of the form of the model to the theoretical concept, expressing the content of relationships between the variables under consideration and as the accuracy of prediction over the considered time interval (1, T) the observed values of the variable 
equation 
... In general, at the stage of substantiating the econometric model, researchers may face the problem of choosing the most preferable composition of independent factors among a number of alternative options.
Can be distinguished two main approaches to solving this problem:
first
involves an a priori (before building a model) study of the nature and strength of relationships between the variables under consideration, according to the results of which factors that are most significant in their direct influence on the dependent variable are included in the model 
.
And, conversely, factors that are either insignificant from the point of view of the strength of their influence on the variable are excluded from the model 
,
or their strong influence on it can be interpreted as induced by relationships with other exogenous variables;
second
the approach to the selection of independent factors - it can be called a posteriori - involves initially including in the model all factors selected on the basis of meaningful analysis. Clarification of their composition in this case is based on the analysis of the quality characteristics of the constructed model, one of the groups of which are indicators expressing the strength of the influence of each factor on the dependent variable 
.
The "a priori" approach is based on the following assumptions: 1) the strong influence of the factor on the dependent variable should also be confirmed by certain quantitative characteristics, the most important of which is their paired linear correlation coefficient 
The logic of using the pair correlation coefficient in the selection of significant factors in practice is as follows. If its value is large enough (≥0.5 ÷ 0.6), then we can talk about the presence of a significant
5-pair correlation in selection (0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 linear relationship between variables 
and 
or about a strong enough influence 
on the 
.
The greater the absolute value
paired linear correlation coefficient, the stronger this influence (positive or negative, depending on the sign 
The value of the paired linear correlation coefficient should be calculated taking into account the transformation form 
and
v models. For example, if 
, then the correlation coefficient is determined between 
and 
,
etc.; 2) if two or more factors express the same phenomenon, then, as a rule, there should also be a fairly strong relationship between them. This can be indicated by the value of the paired linear correlation coefficient 
... In practice, the relationship between factors is considered significant if 
... In such situations, it is advisable to exclude one of these factors from the model so that the same cause is not considered twice. It should be noted that the given boundary values (in the first case 0.5 ÷ 0.6, in the second
0.8 ÷ 0.9 are rather arbitrary. In each case, they are set individually. When choosing them, the intuition of the researcher plays an essential role.
It is usually considered: if for a factor 
, then at a large number other sufficiently significant factors, information that the factor contains 
regarding the variability of the variable 
,
can be neglected. Sometimes, on the contrary, if the composition of factors is not too wide and the factor 
expresses a phenomenon that is significant from the point of view of theory, then the researcher, trying not to lose information about the patterns of variability of the variable 
,
can leave it in the model even with a smaller value of the sample pair linear correlation coefficient (0.3 ÷ 0.4). Such selection, based on empiricism and intuition, usually does not take into account the accuracy of the estimate of the sample correlation coefficients, which increases with the increase in the sample. With a fixed sample size the accuracy of estimates of all coefficients is approximately the same. The logic of such selection is more focused on the substantive side of the problem of taking into account the relationships between the variables of the model. Significantly complicating the problem of factor selection is the phenomenon
false
correlations, i.e. large values of paired correlation coefficients can also take place in those cases when the tendencies of the processes under consideration coincided by chance, in the absence of a logically substantiated relationship between them. False correlation can get in the way of building a “correct” model for two reasons. Firstly, factors that are not significant from a substantive point of view and are characterized by significant values of the paired linear correlation coefficient can be accidentally introduced into the model. Secondly, the model can be excluded from the point of view of the impact on 
factors in relation to which the hypothesis that they express the same phenomenon as another factor (factors) already included in this model is erroneously recognized. Among the main reasons for including variables with false correlation in the model are often the unreliability of information used to determine the values of factors at different points in time, difficulties in formalizing factors of a qualitative nature, instability of trends in the variables under consideration, irregular shape the relationship between them, etc.
The main way, adhering to which you can avoid mistakes associated with the concept of "false correlation", associated with a qualitative analysis of the problem aimed at substantiating the content and form of the model adequate to it. At the same time, we can offer some general recommendations, which it is advisable to adhere to following this path: 1) the number of factors included in the model should not be too large. Their increase can minimize its practical value, since in this case the model begins to reflect not the pattern of development against the background of chance, but the chance itself; 2) the simplicity of the model largely guarantees its adequacy, since more complex dependencies are often a priori elusive in a limited time interval, but at the same time they can be approximated by fairly simple functions. In other words, a complex model can express to a greater extent the secondary relationships between variables to the detriment of the main ones.
With the a posteriori approach, the composition of the factors of the econometric model is clarified on the basis of the analysis of the values of a number of qualitative characteristics of its already constructed version. One of the groups of such characteristics, the most important in the selection of factors, is formed by the values Student's test calculated for the coefficients for each of the model factors. Using this criterion, the hypothesis about the significance of the influence of the factor on the dependent variable is tested. The final decision on the advisability of leaving the factor or removing it from the model is made on the basis of the analysis of the entire complex
Thus, for practice, we can offer the following a step-by-step procedure for building the final version of the model on basisa posteriori approach: 1) the initial version of the model includes all the factors selected in the course of a meaningful analysis of the problem. For this option, the values of the estimates of the coefficients of the model, their root-mean-square errors and the values of the Student's criteria are calculated; 2) an insignificant factor is removed from the model, characterized by the smallest value of the observed value of the Student's criterion, (provided that the observed value is not more than the tabular value), and thus form new variant models with the number of factors reduced by one. Note that there may be several insignificant factors in the model. However, you should not delete all of them at the same time. It is possible that the insignificance of most of them is due to the influence of the "worst" of the insignificant factors and at the next step of the calculations these factors will turn out to be significant; 3) the process of selecting factors can be considered complete, when the factors remaining in the model are significant, if the resulting version of the model also satisfies other criteria of its quality, then , the model building process can be considered complete as a whole. Otherwise, it is advisable to try to form another alternative version of the model that differs from the previous one either in the composition of factors or in the form of their relationship with the dependent variable at.
Each of these approaches has advantages and disadvantages. The “a priori” way of selecting factors does not have sufficient validity. He mostly uses "direct" quantitative indicators of the "strength" of relationships between the quantities under consideration and does not fully take into account the peculiarities of the complex influence of independent factors on the variable 
, i.e. peculiar effects of "emergence" such influence. At the same time, the use of the a priori approach often makes it possible to clarify some preliminary alternative options for sets of independent factors, to check the initial assumptions of the model regarding the correct choice of the form of relationships between them.
The “a posteriori” approach to the selection of factors, at first glance, is preferable precisely because the expediency of including each of the factors in the econometric model is determined on the basis of the entire complex of relationships between the variables included in the model. However, when the total number of factors is large enough, there is no guarantee that many insignificant, or even false, relationships between them will not prevail over the main ones. As a result, it may turn out that among the first candidates for exclusion will be "named" the most important, significant from the point of view of influence on the variable y, factors. Therefore, in difficult cases, i.e. in the presence of a large number of factors selected for inclusion in the model at the stage of meaningful analysis, experts recommend combining both approaches - “a priori” and “a posteriori” in the formation of their “optimal” composition.
According to these recommendations, using the methods of “a priori” selection, using the meaningful analysis, alternative variants of the sets of factors included in the model are formed. Further, using the methods of "a posteriori" selection, these sets are refined, the corresponding model variants are compared according to a number of characteristics of their quality. It is assumed that the best of the model variants also contains an “optimal” set of factors. As a result, the procedure for selecting factors in an econometric model turns into an enumeration of a certain set of their acceptable combinations, formed on the basis of the "a priori" approach. Going through various options for the compositions of independent factors, considering possible types their interrelationships with the dependent variable, the researcher also forms various versions (modifications) of the econometric model to describe the processes under consideration. In this case, the problem arises of choosing the "optimal" or the most "rational" among them. Usually this problem is solved on the basis of an analytical comparison of the statistical characteristics of the quality of the constructed variants, which are calculated already at known values estimates of their parameters.
In the general case, the "quality" of an econometric model is assessed by two groups of characteristics. The first group includes indicators, criteria expressing the degree of conformity of the constructed model to the basic laws of the process it describes. In the second, indicators and criteria, to a greater extent assess the accuracy of its approximation of the observed values 
... The criteria of the first group include Student's t test used to assess the significance of the influence of each factor 
per dependent variable 
... The second group of criteria is formed by the criteria widely used in statistics and econometrics. multiple correlation coefficient,
coefficient of determination and Fischer-Snedecor test.
Model quality assessment is the final stage of its development and has two goals:
1) check the conformity of the model to its purpose (research goals);
2) to evaluate the reliability and statistical characteristics of the results obtained during model experiments.
In analytical modeling, the reliability of the results is determined by two main factors:
1) the correct choice of the mathematical apparatus used to describe the system under study;
2) a methodological error inherent in this mathematical method.
At simulation modeling the reliability of the results is influenced by a number of additional factors, the main of which are:
Modeling random factors based on the use of midrange sensors, which can introduce "distortions" in the behavior of the model;
The presence of a non-stationary mode of operation of the model;
Use of several different types of mathematical methods within one model;
Dependence of the simulation results on the experiment plan;
The need to synchronize the work of individual components of the model;
Having a workload model whose quality depends, in turn, on the same factors.
The suitability of a simulation model for solving research problems is characterized by the extent to which it possesses the so-called target properties. The main ones are:
Adequacy;
Stability;
Sensitivity.
Assessment of the adequacy of the model. In the general case, adequacy is understood as the degree of conformity of the model to that real phenomenon or object for the description of which it is being constructed. The adequacy of the model is determined by the degree of its correspondence not so much to the real object as to the objectives of the study.
One of the ways to substantiate the adequacy of the developed model is to use mathematical statistics... The essence of these methods is to test the hypothesis put forward (in this case, about the adequacy of the model) on the basis of some statistical criteria.
The evaluation procedure is based on a comparison of measurements on real system and the results of experiments on the model and can be carried out in a variety of ways. The most common ones are:
By the average values of the model and system responses;
According to the variances of deviations of the model responses from the average value of the system responses;
By maximum value relative deviations of the model responses from the system responses.
Assessment of the stability of the model. The robustness of the model is its ability to remain adequate when examining the efficiency of the system over the entire possible range of workloads, as well as when making changes to the configuration of the system. The developer is forced to resort to "case-by-case" methods, partial tests, and common sense. A posteriori check is often helpful. It consists in comparing simulation results and measurement results on the system after making changes to it. If the simulation results are acceptable, confidence in the robustness of the model increases.
The closer the structure of the model to the structure of the system and the higher the degree of detail, the more stable the model. The stability of the simulation results can also be assessed by the methods of mathematical statistics.
Estimating the sensitivity of the model. Quite often, the problem arises of assessing the sensitivity of the model to changes in the parameters of the workload and the internal parameters of the system itself.
This assessment is carried out for each parameter separately. It is based on the fact that the range of possible parameter changes is usually known. One of the simplest and most common assessment procedures is as follows.
1) the value of the relative average increment of the parameter is calculated:
2) a couple of model experiments are carried out with the values, and the average fixed values of the remaining parameters. Model response values are determined 
and 
;
3) its relative increment of the observed variable is calculated:
As a result, for the ith parameter, the models have a pair of values that characterize the sensitivity of the model for this parameter.
Similarly, pairs are formed for the rest of the parameters of the model, which form a set.
The data obtained in assessing the sensitivity of the model can be used, in particular, when planning experiments: more attention should be paid to those parameters for which the model is more sensitive.
Calibration of the model. If, as a result of the assessment of the quality of the model, it turned out that its target properties do not satisfy the developer, it is necessary to calibrate it, that is, to correct it in order to bring it into line with the requirements.
Typically, the calibration process is iterative and consists of three main steps:
1) global changes in the model (for example, the introduction of new processes, changes in event types, etc.);
2) local changes(in particular, changes in some distribution laws of simulated random variables);
3) change of special parameters, called calibration.
It is advisable to combine the assessment of the target properties of the simulation model and its calibration into a single process.
The calibration procedure consists of three steps, each of which is iterative (Figure 1.11).
Step 1. Comparison of the output distributions.
The goal is to assess the adequacy of the MI. Comparison criteria may vary. In particular, the magnitude of the difference between the mean values of the model and system responses can be used. The elimination of differences in this step is based on making global changes.
Step 2. Balancing the model.
The main task is to assess the stability and sensitivity of the model. Based on its results, as a rule, local changes are made (but global ones are also possible).
Step 3. Optimization of the model.
The purpose of this stage is to ensure the required accuracy of the results. Three main areas of work are possible here: additional verification of the quality of the sensors random numbers; reducing the effect of the transient regime; application of special methods to reduce variance.
Formally, the quality of a model is determined by its adequacy and accuracy. These properties are investigated based on the analysis of a number of residuals (deviations of the calculated values from the actual ones):
At the same time, adequacy is a more important component of quality, but first we will consider the characteristics of the accuracy and normality of a number of residuals, since some of them are used in calculating various criteria for adequacy.
Accuracy characteristics
Accuracy is understood as the magnitude of random errors. Comparative analysis of accuracy makes sense only for adequate models: among them, the model with lower values of accuracy characteristics is recognized as the best, which include:
The maximum error corresponds to the maximum deviation of the calculated values from the actual ones;
Average absolute error
shows how much the actual values deviate from the model on average;
Average relative error
;
Residual variance
Root mean square error
.
(72)
The root mean square error is the most commonly used measure of accuracy (due to its relationship to residual variance, which plays a central role in regression analysis). The mean square error is always somewhat more value mean absolute error, but they have a similar meaning - they characterize the average distance of the calculated model values from the actual initial data. Typically, the accuracy of the model is considered satisfactory if the following condition is met:
.
(73)
The characteristics of accuracy can also include a multiple coefficient of determination
, (74) characterizing the proportion of the variance of the dependent variable explained by regression and the multiple correlation coefficient (correlation index):
Checking the adequacy of the model
Significance check is carried out on the basis of t- Student's criterion, i.e. the hypothesis is tested that the parameter measuring the connection is equal to zero.
The average parameter error is:
, (76) and for the parameter:
.
(77)
Calculated values t- criteria are calculated by the formula:
(78) A parameter is considered significant if. The value is determined by the formula STUDENT OBR.2
X (0.95; 46) with the number of degrees of freedom and with probability ( P = 1-) For and. Therefore, in this example, the parameters are significant.
The parameter lies within ;,
and the parameter is;.
The significance of the regression equation as a whole is determined using F- Fisher's criterion:
(79)
Calculated value F is compared with the critical one for the number of degrees of freedom at a given significance level (for example,), where,.
If, then the equation is considered significant.
Another approach to determining the values of the parameters of the paired regression equation and assessing the significance is to refer to the mode "REGRESSION" EXCEL. It should be noted that the calculation results shown in Tables 7-9 were obtained with less time and completely coincide with the results of “manual” counting.
Checking for the presence or absence of systematic error
Checking the zero mean property.
The average value of a number of residues is calculated (Table 10):
If it is close to zero, then it is considered that the model does not contain a systematic error and is adequate according to the criterion of zero mean; otherwise, the model is inadequate according to this criterion. If the mean error is not exactly zero, then to determine the degree of its proximity to zero, we use t- Student's criterion. The calculated value of the criterion is calculated by the formula
and is compared with the critical one. If inequality holds, then the model is inadequate for this criterion.
Checking the randomness of a number of residuals.
It is carried out according to the batch method. A series is a sequence of consecutive values of a series of residues for which the difference has the same sign, where is the median of a series of residues.
If the model reflects well the investigated dependence, then it often crosses the line of the graph of the initial data, and then there are many series, and their length is small. Otherwise, there are few series and some of them include a large number of members.
Errors with the same signs located in a row are considered as a series. Next, the number of series and the length of the maximum of them are calculated. The obtained values are compared with the critical ones.
(82) (83) (square brackets round down to the nearest integer).
If the system of inequalities is satisfied:
, (84) then the model is recognized as adequate according to the criterion of randomness, if at least one of the inequalities is violated, then the model is recognized as inadequate according to this criterion.
Checking the independence of consecutive residuals.
It is the most important criterion for the adequacy of the model and is carried out using the Durbin-Watson coefficient:
... (85) For series with a close relationship between successive residual values, the value is close to zero, which indicates that the regular component is not fully reflected in the model and the regularity is partially inherent in a number of residuals, i.e. the model is inadequate to the original process.
If successive residuals are independent, then it is close to 2. This indicates a good quality of the model and clean filtration of the regular component.
With negative autocorrelation of residuals (strictly periodic alternation of their signs), it is close to 4.
To check the significance of the positive autocorrelation of the residuals, the value is compared from Table. 2 Appendices to the lecture:
If, then an assumption arises about the negative autocorrelation of the residuals, and then not, but similar conclusions are drawn with the critical values.
In addition to the most obvious qualities required to become a model (height, size, beauty and health), there are also less obvious, less tangible factors that are also taken into account. They are much more subjective, and the opinion of one modeling agency may differ significantly from the opinion of another. If you become a professional model who regularly attends auditions, you will find that this applies equally to potential employers.
Photogenic
Everyone who has photographed knows that some people turn out well in the picture, while others do not. How photogenic you are depends on how the lens "sees" you.
Your success as a model depends on how you get in photography - how you actually look is not the most important thing.
If you really have this quality, it will definitely show up in the picture. Your mom can take pictures that are very flattering to you with a cheap camera. You can spend as little as a dollar in an instant photo booth and get a strip of great shots.
Chameleon factor
“When I first started my modeling career, my agent said:“ You look great and you will succeed, but you will never become a 'super model', ”the model recalls. Jody Kelly... "He said that because I was like a chameleon and looked different in every photo."
“In order to become a 'supermodel' in the nineties,” she adds, “you had to look so that when you look at your photographs you can say,“ This Cindy Crawford, and here she is again, and again. "
Lucky Jody Kelly in the other, models with a “chameleon factor” tend to have a longer career. And she confirmed it personal example having worked as a professional model for over 10 years.
Patience
One of the most important qualities the model is patience. It sounds obvious, but it's not a commonplace thing.
“Some models do not require this quality, because they have daily photo sessions, they are busy with work,” comments one makeup artist. “Some of them no longer enjoy their work. They work for long hours, and this is bad for themselves and for the quality of their photos. "
“If you’re not patient, just become a model,” emphasizes Sam, Engagement Manager. Ford Models.“You just won't survive. You won't last long. And in general you will not succeed. You have to have patience in this business. "
“It seems to me that these days people, especially the younger generation, have begun to forget that not everything comes at once,” he adds. You can bypass all potential customers in the city and be unemployed for six months ... and suddenly, once, and you get it. This is what true patience is. "
“This means, in the end, to be in the right place and in right time- finally, after a million times you've been out of place and at the wrong time! "
Even after you've got the job, you still have to wait. “Hairstyling and makeup usually takes 3-4 hours to prepare for a shoot. Once you are fully dressed and in character, you are in for a photo shoot and it’s a truly captivating moment. But, believe me, the way to it is not at all! "
The story of a photographer for the Avon catalog
Model Jody Kelly worked on a photo, and the way she did it clearly illustrates what qualities a good model should have.
An order was made for a photo in winter clothes outdoors for the catalog Avon... The problem was that, according to the plan, the filming was usually carried out six months before the release of the catalog. It was just the height of July - the peak of the terrible heat.
Knowing that he would have to shoot in difficult conditions, the photographer chose a model based on three criteria:
1. Her appearance should have been the same as it is accepted in Avon catalogs - a charming, happy smile is obligatory.
2; She never complains about difficulty and gives herself entirely to work.
3. She does not sweat as much as ordinary people... “Jody is insanely dry! Eric jokes. “However, it was precisely this quality that allowed her to receive this order.”
For filming working group chose a nursery Christmas trees in New Jersey, and the photographer, model, makeup artist and stylist went there in a rented refrigerator van.
Snow was sprayed on the branches of the trees. A vast canvas stretched over the model, like a cloud, turned direct sunlight into scattered rays. Jody was wearing a coat, a hat, mittens and a scarf, and it was over 30 degrees Celsius.
All members of the group simply drenched in sweat, and often had to interrupt to drink water. All this time she behaved like a staunch fighter, never once complaining about the situation.
“I had nothing to complain about,” Jody comments on the harsh shooting conditions. - All my life I dreamed of being a model. And so I got this job, and I'm still getting paid to make my dream come true! Even in spite of the weather, I felt happy. "
Gracefulness
"The gracefulness of movement must come from within," explains one representative of the modeling business. “This is something that you were either born with or not. It cannot be sharpened with ballet, sports and other activities. You either have it or you don’t, and this is clearly seen in the photograph ”.
Without complexes
In this business, you can't be overly humble about your body. “You have to feel absolutely comfortable whatever you do with your body,” says the model. Jody Kelly... - During the performance on the podium, you need to get dressed and undress as quickly as possible, and there is no time left for something else to hide behind.
When filming for a magazine, she adds, they will pin something on you, they will touch you with their hands, apply makeup on your face, neck, body. You will be pulled, pulled by your hair, and tied up. At the same time, no one encroaches on you and does not overstep the boundaries, just all these manipulations are necessary and everything must be done very quickly. "
Self confidence
Self-confidence and a sense of comfort are hidden qualities that a model desperately needs. If you're feeling awkward, shy, or awkward, or just upset that a pimple has popped on your face, your lack of self-confidence will show up in the photo.
Openness
Whatever people you work with in the modeling business, you shouldn't be prejudiced against them. You must be able to work successfully with anyone from all walks of life.
"Big" photo
Finally, you need to come to the understanding that modeling is “more than just photography,” and that only experience can give you that. This is difficult for a young model who is under the spell of romance and who has seen enough bad films and read too many tabloid articles.
At the same time, the model does not realize that the creative plan may have been developed over several weeks and that all competent authorities have already approved each of its points, as well as the fact that decisions were made by much more experienced people than she.
It gets sad when you see a model trying to change her makeup, hairstyle or outfit because she thinks she knows how to look better.
If you don’t trust the opinion of the creative team, or you try to explain to your employer what is best for them, you will not only provoke their anger, but you will also lose everything that their creative thought can give you. This will end up with your portfolio filled with one-sided, uninteresting photos.
It should always be remembered that you are just one of the parts that make up one “big” photograph. It is necessary to be sensitive to what other people around you are doing, since they have their own, possibly different from yours, view of postures, facial expressions, clothing and light. Thus, it is absolutely necessary to comprehend and understand their creative intent together in order to realize what they expect from you.
