The use of smart devices in the digital reform of physical education and its impact on learning outcomes

The user modeling process is shown in Figure 1.

Figure 1.

Schematic diagram of the user modeling process

The sports user model includes a model of the user’s basic physical characteristics and a model of the user’s rating matrix for the program. The basic user characteristics model includes basic attribute information and interests of the user. Here the user keywords are extracted, each keyword is a basic information of the user, and the basic information of the user is age, gender, BMI, sports category of interest, etc., and the user model is represented as shown below: (1) Uu→={(ku1,wu1),(ku2,wu2),(ku3,wu3),⋯,(kun,wun)} where Uu→ denotes the feature vector of user u, kun denotes the nth keyword of user u, and wun denotes the weight of the nth keyword of user u.

According to the sports movement itself has the sports quality can be divided into strength, speed, flexibility, endurance, agility five major sports quality. If the sport has both upper limb and lower limb movement, the keyword of the object is set to 1, then the trunk movement is set to 0. If the sport can have the three major sports qualities of strength, speed, flexibility, their weights are set to 1, then the two major sports qualities of endurance and agility are set to 0. Eventually, the model of the sport is the following equation: (2) Iu→={(pu1,wu1),(pu2,wu2),(pu3,wu4),…,(pu7,wu7),(pu8,wu8)}

Where pun represents the nnd keyword, wun represents the weight of the nth keyword, the weight of the keyword is either 0 or 1, that is, whether the sport has this keyword.

The CB algorithm establishes a relationship between the sport and the user based on the user’s behavior in the system [30-31], and the principle of the content-based recommendation algorithm is shown in Figure 2.

Figure 2.

Schematic diagram of CB recommendation algorithm

The content based recommendation algorithm is modeled in the following equation: (3) sim(A,B)=∑i=1nAi*Bi∑i=1n(Ai)2*∑i=1n(Bi)2

where A_i denotes the degree of preference of the user for the type of sport, and B_i denotes the type of sport to which each sport belongs, and this belonging relationship is non-zero that is one.

Finally, based on the similarity of the sports items in descending order, the Top-N can be selected by the user.

In this section, the constructed recommendation model is loaded into the smart device in the smart classroom of M University, aiming to explore whether the model can provide accurate personalized recommendations for students and provide an experimental basis for the following analysis of the impact of accurate sports recommendations on students’ sports level.

The smart device categorizes students based on three dimensions: physical fitness, sports skills, and interest in sports learning, in order to provide precise personalized sports course recommendations for different students.Figure 3 shows the results of the cluster analysis of students using intelligent devices.Table 1 displays the precise description of the clustered groups.

Figure 3.

Intelligent equipment analysis of students’ clustering

Combined with the chart, it can be seen that the intelligent device classifies students into four clusters, namely, “comprehensive development”, “skill specialization”, “interest-driven” and “potential to be developed”. Which are represented by green, purple, orange and blue colors in the figure respectively. Taking the “all-round development type” students as an example, these students have excellent physical fitness, with all physical fitness indicators reaching or exceeding the average level, have mastered a variety of sports skills, and have a strong interest in sports activities. This category makes up 12% of the total number of students surveyed.

In this paper, four students from different cluster groups were selected based on the results of the above cluster analysis. Twenty sports teaching sports are included in the smart device, which are Tai Chi, Rope Skipping, Swimming, Football, Basketball, Volleyball, Table Tennis, Badminton, Tennis, Gymnastics, Jogging, Yoga, Dance, Taekwondo, Karate, Judo, Wrestling, Boxing, Rock Climbing, and Wushu, which are recorded as sports 1-20 in order. Through the content-based recommender algorithm, the weighted ratings of each student are calculated for each of the 20 sports. To recommend the Top-5 teaching sports that have the highest ratings for them.Figure 4 shows the weighted ratings of four teaching sports.

Figure 4.

The weighted scores of the four students’ sports teaching sports

As can be seen from the figure, the recommendation model of this paper for different types of students shows significant differences in the weighted ratings of different sports. Taking student 1 as an example, this student belongs to the comprehensive development type, and the five types of sports with the highest weighted scores are Tai Chi, badminton, soccer, taekwondo, and rock climbing, with weighted scores ranging from 95.63 to 97.65, and these five types of sports include aerobic, anaerobic, flexibility, and strength training, which are the Top-5 recommended by this model for student 1.

Smart classroom teaching evaluation cannot be separated from the support of information technology and big data, which collects the dynamic movement data generated in the process of students’ learning and analyzes them in combination with other external factors, so as to better evaluate students’ learning effects.

The smart teaching evaluation method adopted in this study is to collect the visualized data generated in the learning process or learning results of students through Super Star Learning Pass and Smart Physical Teacher APP, to comprehensively understand the learning situation of students and realize the evaluation of the whole process of students’ smart learning. The evaluation of smart classroom teaching is shown in Figure 5. Smart classroom teaching relies on Super Star Learning Pass and Smart Physical Education Classroom APP to achieve diversified evaluation. The intelligent physical fitness classroom can use modular analysis technology to analyze data sources from different intelligent physical fitness equipment and produce visualized analysis results. The smart classroom teaching evaluation of pre-test questions, cooperative discussions, teaching tasks, classroom exercises, and after-class practice assignments can be used as the evaluation content and evaluation methods, and the evaluation process covers the whole process of course teaching.

Figure 5.

Intelligent classroom teaching evaluation

In this paper, we designed the Physical Education Learning Interest Scale for Higher Education Students, which consists of 20 items covering four different dimensions, of which the dimension of situational interest arousal (A) corresponds to questions 1-5. The dimension of maintaining situational interest (B) corresponds to questions 6-10. The emergent dimension of individual interest (C) corresponds to questions 11-15. The maturation dimension of individual interest (D) corresponds to questions 16-20.

The questionnaire took the form of “Strongly Disagree”, “Disagree”, “Don’t necessarily”, “Agree”, “Strongly Agree” on a scale of 1 (Strongly Disagree) to 5 (Strongly Agree) as a measure of the students’ interest in learning physical education, and the higher the final score, the better their interest in learning physical education.

The statistical results of the test indicators of students’ physical fitness, sports skills, and interest in physical education before and after the test were systematically collected and analyzed using the spss23 application software to compare the differences between the experimental group and the control group, and ultimately concluded.

In this study, the experimental method was used to conduct empirical research, and the second-year students of the School of Physical Education at University M were divided into two groups of 50 students each. Those who were taught physical education in a smart classroom were called the experimental group (T), while those who were taught conventional physical education were called the control group (CK). Before the beginning of the experiment, the physical fitness, sports skills, interest in physical education and other indicators of all students participating in the teaching experiment were measured before the experiment to ensure that the level of the students participating in the teaching experiment was at the same level.

After the completion of the experiment, the data of the experimental group and the control group obtained after the experiment were summarized using the controlled teaching conditions.

Regression models are divided into linear regression models and nonlinear regression models, and linear regression is further divided into univariate linear regression and multivariate linear regression [32-33]. The regression model is as follows: (4) y=β0+β1x1+⋯+βpxp+ε

Where β₀,β₁,β₂,⋯,β_p is the p + 1 unknown parameter, β₀ is known as the regression constant, β₁,β₂,⋯,β_p is known as the regression coefficient, y is known as the explanatory variable, and x₁,x₂,⋯,x_p is p a general variable that can be precisely measured and controlled, known as the explanatory variable. When p = 1 this is the one-way linear regression model, p ≥ 2 it is the multiple linear regression model, and ε it is the random error.

As with univariate linear regression, for the random error term we often assume E(ε) = 0, Var(ε) = σ². called the theoretical regression equation: (5) E(y)=β0+β1x1+⋯+βpxp

If n set of observations (x_i,x_i2,⋯,x_ip;y_i), i = 1,2,⋯,n is obtained, the linear regression model Eq. (6) is: (6) { y1=β0+β1x11+β2x12+⋯+βpx1p+ε1y2=β0+β1x21+β2x22+⋯+βpx2p+ε2⋯⋯⋯⋅yn=β0+β1xn1+β2xn2+⋯+βpxnp+εn is written in matrix form as: (7) y=Xβ+ε

X is a n × (p + 1) matrix and call X the regression design matrix. In order to facilitate the parameter estimation of the model, there are some basic assumptions about equation formula (7) as follows: 1)

The explanatory variables x₁,x₂,⋯,x_p are deterministic and X are full rank matrices.

For the matrix form of multiple linear regression Eq. (7), this condition is: (10) ε~N(0,σ2In)

From the above assumptions and the properties of the multivariate normal distribution, it is known that the random vector y obeys a n-dimensional normal distribution: (11) y~N(Xβ,σ2In)

The “linearity” of a linear regression model is with respect to the unknown parameter β_i (i = 0,1,2,⋯,p). Linearity for the regression explanatory variables is non-essential because the explanatory variables are non-linear and can often be made linear by substituting variables.

This section explores the impact of a smart classroom teaching model for physical education that involves smart devices on students’ learning outcomes in physical education through regression modeling. In this study, teacher qualification, student interest, teaching assessment, teaching content, and teaching mode were used as control variables, online learning activities, interactive discussions, personalized exercise recommendations, student fitness monitoring, and teaching feedback provided by smart devices were used as independent variables, and learning effect was used as dependent variable. The results of the regression analysis between smart devices and learning effects are shown in Table 3.

The regression results show that student interest, teaching assessment and teaching mode have a significant positive effect on the learning effect in the process of physical education teaching, with correlation coefficients of 0.118, 0.237 and 0.156, respectively, and p-values of less than 0.001, i.e., for every unit increase in student interest, teaching assessment and teaching mode, the learning effect of physical education can be increased accordingly by 0.118, 0.237, 0.156 and passes the significant test at the 0.001 level. In addition, a variety of influencing factors intervened by intelligent devices in the teaching process have a facilitating effect on the physical education learning effect at the 0.001 level, among which the personalized recommendation provided by intelligent devices has the most obvious facilitating effect, and when it is increased by 1 unit, the learning effect rises by 0.347. The regression results of the present study show that the R-square is 0.493, which indicates that the above variables have 49.3% explanatory explanation for the dependent variable’s ability with good effect.