Paper Title (use style: paper title)

III. Method for determining informative parameters

The determination of informative parameters for predicting the predicted parameter can be carried out by various methods. One such method is the backward elimination method.
This method starts by including all possible parameters in the regression model and then excluding parameters that do not significantly contribute to explaining the variation in the predicted parameter. In each iteration, the regression model is rebuilt, excluding one parameter, and a statistical analysis of the significance of the remaining parameters is performed. If the remaining parameters remain significant, then the excluded parameter is not informative and can be excluded from further analysis.
There are also other methods for determining informative parameters, such as the method of multiple regression, the method of principal components, the method of independent components, etc. Each of these methods has its advantages and disadvantages and can be applied depending on the specific task and data.
The methodology for determining a set of informative parameters for individual forecasting by methods of pattern recognition theory may include various stages, such as:
1. Collection and preparation of data. At this stage, it is necessary to collect data and prepare it for analysis. This may include data cleaning from outliers and missing values, data aggregation, and other manipulations.
2. Selection of informative parameters. At this stage, it is necessary to determine which parameters are most informative for predicting the target variable. For this, various methods can be used, including correlation and regression analysis, principal component methods, recursive feature elimination, and others.
3. Correction of correlation coefficients. If the selected parameters are highly correlated with each other, it is necessary to adjust the correlation coefficients to avoid model instability and incorrect results.
4. Choice of the most informative parameter. At this stage, it is necessary to choose the most informative parameter for predicting the target variable. This can be done through various methods, including ranking methods and machine learning methods.
5. Building a model. At this stage, it is necessary to build a forecasting model based on the selected informative parameters. This can be done using various methods including linear regression, decision trees, neural networks, and others.
6. Assessment of the quality of the model. At this stage, it is necessary to evaluate the quality of the constructed model using various metrics, such as the coefficient of determination, root mean square error, and others.
Thus, the methodology for determining a set of informative parameters for individual forecasting by methods of pattern recognition theory includes several stages, starting with the choice of informative parameters and ending with assessing the quality of the constructed model.
As a result of the implementation of all stages of the methodology for determining informative parameters for predicting the indicators of the state of sports readiness of an athlete, it is possible to obtain a model that allows more accurately predicting changes in the indicators of sports readiness of an athlete and, thus, more effectively plan his training process.
Analysis of the results of training is a rather laborious task. Therefore, computer and statistical calculation programs (STATISTICA, MathCAD, Mathematica, PolyAnalyst, MiniTab, etc.) are used for its implementation [7-9]. To simplify the calculations, it is recommended to bring nonlinear constraints to linear ones, i.e. perform a linearization operation (splitting the nonlinear characteristic into linear sections). Recommended criteria for making the informative parameter significant:
- the correlation coefficient between the predicted and informative parameters is not less than 0.3;
- the correlation coefficients between the informative parameter and other informative parameters are not more than 0.7.
As an example, consider the selection of the teams “Olympia” and “Sevinch”, recommended for international competitions. The sample size was 28 female athletes. Such a volume was previously justified and tested in practice [10-13]. The predicted parameter y is hit by a range. As informative parameters, [13] are considered:
- x1 - running for 30 meters, [s];
- x2 - five-time jump, [m];
- x3 - running 7x50 m, [min, s];
- x4 - ball juggling, [amount].
Determining the degree of connection between the predicted parameter and informative parameters is one of the main tasks of the study. To do this, an analysis of the correlation between these parameters is carried out, general correlation coefficients are estimated, and regression curves are built. Based on the obtained results, it is possible to draw conclusions about the influence of informative parameters on the predicted parameter and use them to optimize the training process and increase the sports readiness of athletes. In table. 1 shows the values of the parameters of the sample athletes after the training experiment.

1. 
   Calculated parameters of athletes in the sample after the training experiment
   

№