Determination of the optimal measurement time to improve the accuracy of electrical conductivity data processing

. In this paper, the possibility of improving the accuracy of the assessment of primary and secondary diagnostic indicators is considered. The electrical conductivity of distilled water was measured with subsequent processing of descriptive and correlation analysis methods using the Statistica computer program.


Introduction
Water is an important component that determines the properties and structure of a huge variety of inanimate and living nature. The uniqueness of this component lies in the fact that it is able to respond to any impact. For example, the influence of a given electromagnetic field on the physical and chemical properties of water was studied in the works [1][2][3][4][5]. In works [6][7][8][9], as well as in works [10][11][12][13][14][15], studies of the properties of aqueous solutions after various effects of physical nature were carried out.
Statistical evaluation of processes is currently one of the most important stages in the analysis of measurements. This is due to the fact that an incorrect assessment reduces the accuracy of the assessment of primary and secondary diagnostic indicators characterizing a set of signs with which it is possible to determine the condition of the object of control. Improving the accuracy of the evaluation of indicators and, consequently, the reliability of the analysis, mainly depends on the effectiveness of the primary processing methods used [16]. The purpose of the research is to increase the effectiveness of data processing by determining the optimal study time.

Initial data, progress of work and processing of results
In the work, the results of measurements of the electrical conductivity of water were processed using a professional water quality meter WMM 97 at a temperature of 21 ° C (the pH of the medium corresponded to GOST 58144 5.0-6.6). The study of the electrical conductivity of distilled water was carried out for 1 hour, 3 hours, 1 day, 3 days (electrical conductivity corresponded to GOST 58144, no more than 0.43 mSm / m). A sensor with an electrode is placed in a cell with water, which is left for a given time. The device takes readings of the measured physico-chemical quantity every second. We have chosen such an indicator as electrical conductivity, because it is one of the fundamental ones for industry and the scientific world from the side of physics and chemistry. The results, at a given study time, were processed using the computer program Statistica [17,18].
The maximum and minimum values of samples, median, mean value of the sample, asymmetry, kurtosis, standard error of asymmetry and kurtosis were obtained.   At the initial stage of the primary analysis of experimental data, the numerical characteristics of descriptive statistics were investigated. The sample arithmetic mean, median, lower and upper quartile (percentiles 25% and percentiles 75%) and quartile range (interquartile range or percentiles) were obtained. Using these characteristics, it is possible to construct (Fig. 1-4) and analyze boxplot diagrams, and then check the distribution of the sample for compliance with the law of normal distribution.         Scattering diagrams were also constructed in order to clearly show the relationship between electrical conductivity and constant temperature over time, which can be described by a locally linear model of locally weighted regression (lowessa), the corresponding dependencies are shown in Fig. 13-16.

Conclusion
As a result of the work performed, probabilistic graphs, box diagrams, as well as the dependences of electrical conductivity on the time of the process under normal conditions were constructed. The study was conducted in order to improve the efficiency of data processing by determining the optimal time of the study using the computer program Statistica. The data obtained are close to the GOST data. But it can also be concluded that as the measurement time increases, the probability of errors and outliers increases. Therefore, based on experimental data, we can say that the optimal time for studying the electrical conductivity of aqueous solutions can be considered a three-hour study. Three-hour measurements are relevant if structures such as cluster types and proton conductivity are not taken into account.