Establishment and analysis of urban talent attraction evaluation model

. Nowadays, the attraction to talents has become increasingly prodominent in determing the future development prospect of cities. How to quantify the attractiveness of cities to talents is an important issue in evaluating the comprehensive competitiveness of cities. This paper takes Xi ' an and other emerging first-tier cities in China as the main research object, and classifies the indicators that affect the attractiveness of urban talents into five specific categories. A set of evaluation models are constructed by using entropy weight method, information entropy method and TOPSIS algorithm. The different indicators that measure the level of urban development are normalized and weighted to obtain the attractiveness index of urban talents to describe the attractiveness of talents in cities with similar development levels. To prove the rationality of the model, we apply this evaluation model to different cities and test its stability.


Introduction 1.Background
Talent is a key factor in the development of a city, which is of great significance to the city's economic growth, industrial upgrading and comprehensive competitiveness.At the same time, talents can also enhance the vitality and innovation of a city.In current China, with the gradual disappearance of demographic dividend, the importance of talent is increasingly prominent [1] .Therefore, more and more cities, especially the new first-tier cities which are in the stage of rapid economic development, have introduced preferential policies to attract talents to improve their comprehensive competitiveness.
Under the background of competing for talents in various cities, how to evaluate the attraction of a city to talents and how to formulate appropriate policies to make a city more attractive to talents have become issues of great concern to urban managers.
The influencing factors of talent attraction involve economics, psychology and many other fields, and are closely related to urban policies.To evaluate the talent attraction of cities, it is necessary to modify and model these influencing factors so that they can be quantitatively compared.This is also valuable for urban managers to formulate talent attraction policies.

Construction of the talent attraction evaluation model
Considering the competition with Xi ' an in terms of talent attraction is mainly other new first-tier cities, we first collected the relevant data of eight new first-tier cities including Xi ' an.Subsequently, we apply three different algorithms to weight the data, and then combine the weights to multiply the 12 indicators after normalization, and get the talent attraction score of each city, so as to evaluate the health status of talent attraction in Xi 'an.

Data collection and collation
We collect specific data of eight new first-tier cities under 12 indicators in five categories in 2020 [2] , the indicators we selected and the aspects they belong to are shown in Table 1: To simplify the evaluation model and quantify the attractiveness of urban talents, We assume that a city ' s talent attraction is only related to the five aspects mentioned before and each aspect is determined only by the indicators presented above.

Evaluation model establishment
In this section, we will introduce the ideas of three different algorithms: the TOPSIS Algorithm, the Entropy Algorithm and the Information Weight Model.Then, we will integrate the three algorithms to construct our evaluation model.

TOPSIS Algorithm for determining weights
The idea of TOPSIS algorithm is to construct an optimal vector and a worst vector, and to judge the pros and cons of the scheme by the distance between the index and the optimal vector and the worst vector. [3,4]The closer the index is to the optimal vector and the farther it is to the worst vector, the better the scheme is.

Entropy Algorithm for determining weight
In information theory, entropy is a measure of uncertainty.The greater the uncertainty is, the greater the entropy is, the more information it contains, the smaller the uncertainty is, the smaller the entropy is, and the less information it contains. [5]ccording to the characteristics of entropy, the dispersion degree of an index can be determined by calculating the entropy value.The greater the dispersion degree of the index is, the greater the weight of the index to the comprehensive evaluation is.
The specific process of determining data weight by Entropy Algorithm is as follows: Calculate the entropy of indicator j: Calculate the weight of each index: Construct Weight Matrix W:

Information Weight Model for determining weight
Information weight method is also called variation coefficient method.Information weight method is an objective weighting method.The idea is to use the coefficient of variation of data to assign weights.If the coefficient of variation is larger, the amount of information it carries is larger, and thus the weight is larger. [6]he specific process of determining data weight by Information Weight Model is as follows: Obtain calculation matrix Y by normalizing data matrix X: Calculate entropy of index j: Calculate evaluation index weight:

Integrating weights of three models
In order to make the final weight of each index more credible, we consider the index weight calculated by three different models comprehensively, and combine the three models with an average method to obtain the final weight of each index as shown in Table 2. [7] Table 2. Weight calculation result.

Application model for the city of Xi'an
In order to evaluate the health degree of talent attraction in Xi ' an, the optimal value of each index of the selected new first-tier cities is 100, and the 12 indexes of Xi ' an are normalized.
Then the normalized data are multiplied by the weight of each index in the comprehensive weight model, and the results are used as the standard to judge the health degree of talent attraction in Xi ' an.According to the above steps, we can answer the first question.We calculated Xi'an relative to other new front-line cities talent attraction health degree is 75.52 out of 100.

Applicability of evaluation model to different cities
In order to verify the applicability of our evaluation model to different cities, we apply the evaluation model to the selected seven new first-tier cities.Five factors and 12 indicators of these cities are scored and weighted respectively, and the talent attractiveness scores of these cities are obtained.We compare the talent attraction scores of these cities with their actual talent introduction in 2020, and find a basic fit, which verifies the applicability of our model to different cities.

Talent attraction in different new first-tier cities
We apply the model to the selected seven new first-tier cities, and get the scores of seven cities and Xi ' an on the five factors that affect the attractiveness of urban talents.The results are shown in Figure 1.According to the weight calculated by the model, the talent attractiveness scores of these cities are weighted and ranked, and the results are shown in Table 3.In general, we believe that the above ranking results are basically consistent with the talent attraction rankings of these cities in reality, which can prove the applicability of the analysis model constructed by us.

Sensitivity analysis
In the Talent Evaluation Model of this paper, due to the different sources of data collected on the 12 indicators of different cities, it could lead to inaccuracy of the input data.And this may bring some deviations to the results of our model.Therefore, to verify the stability of our Talent Evaluation Model, We change the growth rate of the set indicators.It can be seen from Figure 2 that when the indicator growth rate of each year is 1 %, the difference between the result score and the original value is about 2.5 %, and the line fitting degree is high, showing the robustness of our model. [8]g. 2. Score changes of Xi ' an in ten years before and after adjustment.

Conclusion
The attraction of a city to talents is determined by many aspects.In order to quantitatively evaluate the health of talent attraction in a city, we identified five categories affecting talent attraction and 12 specific indicators.We collected relevant data from the official website.Through TOPSIS Algorithm, Entropy Algorithm and Information Weight Model, we successfully realized the weight distribution and normalization algorithm of 12 indicators, and finally constructed an evaluation model to evaluate the attractiveness of talents in eight new front-line cities.In conclusion, we believe that this evaluation model can better evaluate the status of urban talent attraction, and by analyzing the stability of the model, we prove the applicability and universality of the evaluation model.
In the future, we believe that this evaluation model can be applied to more cities at different stages of development, so as to better evaluate the talent attraction of different cities and further improve the model.

Fig. 1 .
Fig. 1.Comparison of Influencing Factors of Talent Attraction in Different Cities.

Table 3 .
Talent attraction score and rank.