Audit and economical analysis

Table 25

Statistical characteristics of initial data

Change number	Average arithmetic value	An average quadratic deviation	Variation,%	Asymmetry	Experiment	Asymmetry error	Exception error
Y	27.15	2.85	10.50	0.20	-1,16	0.37	0.73
x1	2.77	0.28	10.08	0.36	-0.81	0.37	0.73
x2	92,57	8,70	9.39	0.24	-0.69	0.37	0.73
x3	8,46	0.59	7.00	0.10	-0.52	0.37	0.73
x4	17,77	2.76	15.55	0.72	-0.08	0.37	0.73
x5	31.68	7.28	22.98	0.63	-0,13	0.37	0.73

 

The highest variation in the table is x ₅ ( B = 22.98) but does not exceed 33%. So, the startup information is in one round and can be used for the next account.

On the basis of the highest variation, the data required for correlation analysis can be found in the formulas below.

                                                       

where n - the required amount of information to be selected;

V - variation,%;

t - probability ratios of reliability at P = 0.05 to 1.96;

m - the accuracy of the calculations (allowed in the economic indices up to 5-8%).

Thus, the size of the selection (40 enterprises) is sufficient for correlation analysis.

The primary concern of the subordinate division is that the majority of the data being learned should be grouped around its mean value for each indicator, and that large and smaller objects should be as much as possible. The normal distribution chart is shown in Figure. 


Graph. Graph of normal distribution of data.
An asymmetry index is used to quantify the degree of deviation from the normal distribution.

The asymmetrical Index ( A ) and error ( _m ) calculated according to the following formula:                                                                                   

The Extraction Index ( E ) and its error ( m _e ) are calculated according to the following formula:                                                                     

Symmetric distribution A = 0 . In contrast to zero, there is an asymmetry in the distribution of data. Negative asymmetry suggests that the data is of a high value, while the positive asymmetry is a large number of small values.

In the normal distribution, the extensibility index is E = 0 . If E > 0, then the mean value of the data is formed by a sharp tip formed in a dense concentration. If E < 0 , the curve distribution will be in the form of a finite shape. However, A / _m and e / _m 3 is smaller than the speed of response and ekstsess not reviewed in significance and distribution in accordance with the laws and regulatory information.

in the table A / m _a and E / m _e in any desired relationship, so the initial information is subject to this law.

Modeling of relationships between objective and outcome indicators focuses on the choice of the equivalence equally reflecting the studied link.

Methods for establishing existing links to justify it are: analytical grouping, linear graphs, and so on. is applied.

If all factor indicators are in the linear character of the resulting link, then you can use the following linear function to write that link                               

The function of the linear form of communication between the result and the factor   or logarithmic function.                                                       

The best aspects of the two models is that they ( b ) can be an indicator of economic thought (comment). In the linear model, when coefficients b _i are converted to factor-based units, the absolute expression varies with the degree of logarithmic and percentages, and shows how the resulting cost has changed to unit.

When it is difficult to justify the connection, it is possible to apply different models and compare the results to solve the problem. We use Fisher, which measures the average error of approximation and the size of the coefficients of determinism to verify the compatibility of different models in real connections. Let's talk about this later.

Studying the relationship between the factors being studied and the level of profitability has shown that all the connections we have considered in our example have a straight line character. Therefore, the linear function was used to represent them.

Factor analysis of the korrilyatsion to resolve the issue Shehu M 'Statistics' package design programs . First of all, the starting data matrix (the table) is created. In the first column, the trace number is counted, the value of the endpoint ( Y ) in the second column , and the next digit ( xi ).