An Exploration of the Path of Information Technology to Strengthen the Teaching Quality of Physical Education in Colleges and Universities

Physical education plays a pivotal role in quality education. Nowadays, when quality education is advocated, people must pay enough attention to physical education to make it work better. The goal of physical education in colleges and universities is not only to cultivate students’ sports skills, but more importantly, to promote students’ all-round development and to cultivate healthy lifestyles and positive attitudes towards life [1–3]. However, with the development of society and the progress of science and technology, the traditional physical education model and methods have been unable to meet the needs of contemporary students. At the same time, the quality of physical education teaching in colleges and universities also faces some challenges, such as insufficient teaching resources and uneven teaching level of teachers [4–6]. The rapid development of modern information technology provides a broad application space for physical education teaching in colleges and universities. Modern information technology includes virtual reality technology, intelligent teaching aids, online learning platforms and resources, and data analysis and evaluation technology. These technical means provide rich teaching resources and teaching means for physical education teaching in colleges and universities, broaden the boundaries of teaching, and enhance the flexibility and personalization of teaching [7–9]. Teachers should take the initiative to use information technology to assist teaching and present physical education content in a more intuitive and innovative way, so as to fully mobilize students’ enthusiasm and achieve the expected teaching purpose [10–11]. Information technology has been widely used in all walks of life, producing obvious social and economic benefits. The integration of information technology into the classroom of college sports can create a suitable classroom environment for students, thus enhancing the effectiveness of classroom teaching and changing the drawbacks of traditional classroom teaching methods. In college physical education teaching, teachers should take the initiative to use computer technology, so that the students’ physical fitness can be effectively improved, so as to achieve the purpose of comprehensive development [12–15].

The development of information technology has promoted the improvement of the quality of physical education teaching in colleges and universities. Literature [16] emphasizes the importance of curriculum reform, tries to use the method of network computer server cluster to design the teaching content of physical education professional courses, and verifies the feasibility of the curriculum reform through actual case study, which can maximize the professional function of physical education teaching and better provide basic guarantee for physical education teaching. Literature [17] points out that network teaching is the new trend of education innovation nowadays, and by examining the application of network interactive teaching in physical education network courses to create a new teaching mode of physical education courses, which in turn improves students’ motivation to learn, enhances the interaction and communication between teachers and students in the classroom, and thus strengthens the quality of physical education teaching. Literature [18] proposed a continuous filtering convolutional neural network and designed a physical education teaching intelligent assistance system based on it, and verified the superior performance of the system through experimental tests, which can assess learners’ comprehension, memory and achievement, and also provide personalized training plans to students. Literature [19] based on fuzzy decision tree algorithm constructed a college sports MOOCS teaching model, the evaluation and analysis verified that the proposed teaching model has strong practicality, which can solve the problem that the traditional sports can not meet the needs of “Internet + education”. After exploring the characteristics, functions and dilemmas of physical education teaching in higher vocational colleges and universities, [20] puts forward the proposal of introducing computer network technology in physical education teaching, so as to increase the interest and depth of the process of physical education teaching in schools and thus improve the quality of physical education teaching. Literature [21] from the modern computer technology and college physical education teaching two aspects, designed a computer technology-based physical education teaching model, through the actual teaching experiments to verify the feasibility of the teaching model, can effectively improve the efficiency and quality of physical education teaching. Literature [22] combines artificial intelligence technology with sports, and through the analysis of this new sports teaching mode, it is found that artificial intelligence applied in sports teaching can satisfy the personalization of sports teaching and the intelligence of teaching management, realize the precision of sports teaching, improve the teaching efficiency and classroom effect, and help to improve the students’ physical fitness and academic performance.

In this paper, the basic theory of cluster analysis algorithm is firstly investigated and further explored the data mining based exercise prescription generation technique, which is applied to the screening of exercise programs. Subsequently, the feature information of the students is clustered, and the students are compared with the centroids of each cluster to derive the cluster in which they are located, and then the similarity comparison is performed to derive the students with the highest feature similarity and recommend their exercise prescriptions. Finally, the optimized K-means clustering algorithm was used to cluster and divide the 6000 male and 6000 female test data in the physical education teaching database of a university as the experimental data, followed by the exercise prescription parameter recommendation experiment. And the verification of the effect of physical education teaching is carried out by analyzing the comparison between the experimental group and the control group in terms of special physical fitness, learning interest and total performance.

2

Method

2.1

Cluster analysis algorithm

2.1.1

Cluster analysis

Cluster analysis is different from categorical analysis, which is to cluster data objects based only on their own characteristics without knowing the number of categories and classifications. Since cluster analysis is based on the intrinsic characteristics of the data objects, the number of categories and classification labels do not need to be specified beforehand, so the results may be different for different researchers clustering the same kind of data [23]. Sergios T. in Pattern Recognition (Fourth Edition) defines cluster analysis as follows:

Suppose w = {x₁, x₂, …, x_n } is the sample dataset and c₁, c₂, …, c_k is the partitioned class cluster, where x = (x₁, x_i2, …, x_im)^T, if satisfied: 1)

c_i ≠ ϕ, i = 1, 2, …, k.

2)

$\cup_{i = 1}^{k} c_{i} = w$ .

3)

c_i ∩ c_j = ϕ, i, j = 1, 2, …, k, and i ≠ j.

Where condition (1) ensures that none of the categorical sets obtained from clustering is an empty set, i.e., contains at least one element. Condition (2) ensures that each element in the sample data set can be classified in a different class cluster. Condition (3) restricts each data sample to be classified in only one class cluster, i.e., each categorical set is independent.

In order to carry out reasonable clustering division for data of different shapes, types and sizes, researchers at this stage have proposed a variety of clustering analysis algorithms, but it is a great pity that they have not yet found a sophisticated clustering algorithm that can be universally applied. Therefore, when clustering, it is necessary to choose a most suitable clustering algorithm according to the characteristics of the data itself and the purpose of clustering. The clustering analysis algorithm is shown in Figure 1.

2.1.2

K-means clustering algorithm

The traditional K-means clustering algorithm only needs to adjust the number of clusters as a parameter in advance.

Suppose, the sample dataset to be clustered is w={x₁, x₂,…,x}, the steps to realize the cluster division using K-means clustering algorithm are as follows:

Inputs: sample dataset W, number of clusters k.

Output: k samples divided into class clusters. 1)

In the set W, any k samples c₁, c₂,…,c_k are selected as initial clustering centers.

2)

According to the distance of the remaining samples to c₁, c₂,…,c_k respectively, divide them in the class where the nearest c_i, i=1, 2,…,k is located, and get the set of k initial divided samples.

3)

The mean of the set of k divisions is used as the initial clustering center for the next iteration and the next clustering division is performed.

4)

Iterate repeatedly until the number of iterations is reached or the clustering accuracy is satisfied.

2.1.3

Similarity Measures

1)

Distance measurement

The distance metric is used to measure the distance that individuals exist in space. The greater the distance, the greater the difference between individuals. Assuming that x = { x₁, x₂, ⋯, x_m }, y = { y_i, y₂, ⋯, y_m } are samples in D, denote d(x, y) as the distance between x, y [24]. There are several ways to calculate d(x, y) by distance measure.

(1)

European distance: 1 $d (x, y) = ‖ x - y ‖ = \sqrt{\sum_{i = 1}^{m} {(x_{i} - y_{i})}^{2}}$

(2)

Absolute distance: 2 $d (x, y) = \sum_{i = 1}^{m} | x_{i} - y |$

(3)

Chebyshev distance: 3 $d (x, y) = \max_{i} | x_{i} - x_{j} |$

(4)

Nominal distance: 4 $d (x, y) = {(\sum_{i = 1}^{n} {| x_{i} - x_{j} |}^{m})}^{\frac{1}{m}}$

2)

Correlation coefficient measure

Correlation coefficient is another way to measure the similarity of samples, by calculating the size of its value can determine the degree of correlation between two samples. The following are two methods commonly used to calculate the correlation coefficient:

(1)

Angle cosine method

The cosine of the angle between two vectors in a vector space is used to measure the degree of similarity between two samples, i.e., the cosine of the angle. The closer the value is to 1, the more similar the two samples are to each other: 5 $\cos (x, y) = \frac{\sum_{i = 1}^{m} x_{i} y_{i}}{\sqrt{\sum_{i = 1}^{n} x_{i}^{2}} \sqrt{\sum_{i = 1}^{n} y_{i}^{2}}}$

(2)

Pearson correlation coefficient

Pearson’s correlation coefficient is a statistic used to reflect the degree of similarity between two variables. The range of values is [−1,1], and the larger its absolute value, the stronger the correlation between the two samples. That is: 6 $p (x, y) = \frac{\sum_{i = 1}^{m} (x_{i} - \bar{x}) (y_{i} - \bar{y})}{\sqrt{\sum_{i = 1}^{m} {(x_{i} - \bar{x})}^{2} \sum_{i = 1}^{m} {(y_{i} - \bar{y})}^{2}}}$

2.2

Analysis of Data Mining Based Exercise Prescription Generation Techniques

2.2.1

Exercise prescription data structure

The data structure of exercise prescription mainly contains: type of exercise prescription, exercise program, duration, intensity, frequency and precautions, of which the intensity of exercise is indicated by heart rate or “subjective exercise intensity level”. Exercise prescription has a clear purpose of exercise, so the development of exercise prescription has a strong target, the requirements of the development of exercise prescription process needs to be clear exercise demand, so exercise prescription has a clear classification, exercise prescription is generally divided into aerobic exercise, strength exercise and flexibility exercise three categories. The data structure of exercise prescription under the computer perspective needs to accurately describe the significance of exercise prescription in terms of exercise purpose, process control, effect evaluation, etc. The data structure of exercise program is shown in Table 1.

Table 1.

Movement scheme data structure

Data name	Symbol	Data content	Data type
Prescription type	Type	Aerobic, powerful, flexible	Int
Sports purpose	Effect	The goal required for exercise corresponds to user requirements	Vector
Action site	Target	Body parts that need to be exercised	Vector
Sports	Item	The necessary sports items to be executed	Array
Motor cycle	Period	The time required to achieve the purpose of exercise	Int
Motor intensity	Strength	Best exercise intensity when performing exercise	Int
Motion length	Duration	The best motion time of the exercise	Int
Motion frequency	Frequency	The number of times per week	Float32
Conditional requirement	Amount	Minimum exercise during exercise	Int
Considerations	Attention	The point that needs to be noted during the exercise	--

2.2.2

Analysis of Collaborative Filtering in Exercise Prescription Generation

Selecting appropriate exercise programs is the key to developing exercise prescriptions, and developing targeted exercise prescriptions mainly relies on mining the relationship between exercise programs and exercise demand, while comprehensively considering factors such as students’ physical condition and exercise environment. Collaborative filtering algorithms can uncover potential associations between students and content or commodities in a large amount of student data, so as to establish a mapping relationship between exercise demand and exercise programs [25]. The algorithm is divided into two parts: online collaborative and offline filtering, online collaborative is to find items that are potential for the target students, while offline filtering is to remove items that are not worth recommending, and also needs to filter out items that students have already obtained, so as to discover students’ potential preferred content. Filtering for sports items in sports prescription generation is analyzed as follows: 1)

Content-based collaborative filtering algorithm

Content-based collaborative filtering algorithm (ItemCF), extracts the features that need attention from the filtered objects and uses their features to calculate the similarity, and then uses the preferences of different students to generate a recommendation list. Content-based collaborative filtering algorithms are very popular in personalized applications, calculating similarity based on the popularity of two items or contents, recommending similar items or contents for students, thus mining their potential associations, in addition to being insensitive to the addition of new contents, and being able to expand the data at will. The idea of the algorithm is shown in equation (7): 7 $w_{i j} = \frac{| Q (i) \cap Q (j) |}{\sqrt{| Q (i) ‖ Q (j) |}}$

Where w_ij denotes the similarity of the two items calculated, w_ij denotes the number of people who like item i, so the numerator expresses the number of people who like both items i and j, while the denominator denotes the total number of people who like either item i or j, and the result of the computation expresses the similarity of the popularity of the two items. The students’ preference for the potential target content or goods is further calculated as shown in equation (8): 8 $p (u, i) = \sum_{j \in N (u) \cap s (j, K)} w_{i j} r_{u j}$

In Eq. (8), N(u) is the set of items preferred by the students, S(j, k) is the set of K items similar to item j, and r_uj is the students’ preference for item j, from which the students’ preference for item i is calculated as p(u, i).

2)

Student-based collaborative filtering algorithm

The student-based collaborative filtering algorithm recommendation is based on the inter-student similarity, which searches for a group of students who are similar to the target students, and in this similar group of students, the items that are unknown to the target students but are universally liked are recommended to the target students [26]. Calculation of inter-student similarity can be achieved using Jaccard’s formula as shown in equation (9) below, where student’s interest in the item is used as a behavioral indicator to calculate inter-student similarity: 9 $w_{u v} = \frac{| N (u) \cap N (v) |}{\sqrt{| N (u) | | N (v) |}}$

In Eq. (9), N(u) denotes the set of items of interest to student u. Based on the calculated student similarity, K students who are similar to the target student are formed into a similar student group, and the items of general interest in the similar student group are searched for the unknown potential interest of the target student, as shown in Eq. (10) below, and the calculated result indicates the probability that the target student u is interested in the item i. Then: 10 $p (u, i) = \sum_{v \in s (u, K) \cap N (i)} w_{u v} r_{v i}$

In order to solve the influence of popular items or information on the similarity calculation, the 1

penalty factor $\frac{1}{\log 1 + | N (i) |}$ is added to improve the robustness of the algorithm, so the similarity between students is calculated as follows in equation (11): 11 $w_{u v} = \frac{\sum_{i \in N (u) \cap N (v)} \frac{1}{\log 1 + | N (i) |}}{\sqrt{| N (u) | | N (v) |}}$

3)

Collaborative filtering algorithm based exercise program screening program analysis

Using content-based collaborative filtering algorithm to realize exercise prescription generation, the similarity association between exercise programs targeting specific needs can be mined from students’ needs, and the content-based collaborative filtering screening exercise program is shown in Figure 2. According to the students’ needs, the similarity calculation between exercise programs is established to realize the many-to-many mapping relationship between exercise needs and exercise programs, based on which, this paper can start from the students’ needs, filter the appropriate exercise programs, and then combine with the exercise prescription parameter calculation method to further generate the exercise prescription. However, the classical collaborative filtering algorithm is difficult to meet the analysis of student variability, and in view of the easy scalability of the algorithm, this paper can be improved to achieve personalized exercise prescription generation.

2.3

Design of Personalized Exercise Prescription Recommendation Algorithm

2.3.1

Calculation of exercise prescription recommendations

In this paper, considering that the information of students is mainly data such as age, weight, BMI, body fat percentage, etc., the calculation of similarity is carried out by attribute weighting. 1)

For the student’s age attribute, the similarity between the student’s age and the age of other students in the same category is calculated by formula (12). 12 $S (A_{1}, B_{1}) = \frac{50 - | A_{1} - B_{1} |}{50}$

Where A₁ represents the age of the student and B₁ represents the age of other students in the same category.

2)

For the student’s weight attribute, the similarity of the student’s weight to other students in the same category was calculated using equation (13). 13 $S (A_{2}, B_{2}) = \frac{70 - | A_{2} - B_{2} |}{70}$

Where A₂ represents the weight of the student and B₂ represents the weight values of other students in the individual category.

3)

For the student’s BMI attribute, the similarity of the student’s BMI with other students in the same category was calculated by using equation (14). 14 $S (A_{3}, B_{3}) = 1 - b \times 0.2$

Where A₃ represents the student’s BMI value and A₃ represents the BMI of other students in the same category. b represents the range of difference between the two BMI values, for example, one of the two BMI values is normal and the other one is obese, at this time the range difference is 2, so the similarity is 0.6.

4)

For the student’s body fat percentage attribute, the similarity of the student’s body fat percentage to that of other students in the same category was calculated by using equation (15). 15 $S (A_{4}, B_{4}) = \frac{30 - 100 | A_{4} - B_{4} |}{30}$

Where A₄ represents the student’s body fat percentage and B₄ represents the body fat percentage of other students in the same category.

5)

For the student’s waist-to-hip ratio attribute, the similarity of the student’s waist-to-hip ratio to that of other students in the same category was calculated using equation (16). 16 $S (A_{5}, B_{5}) = \frac{80 - 100 | A_{5} - B_{5} |}{80}$

Where A₅ represents the waist-hip ratio of students and B₅ represents the waist-hip ratio of other students in the same category.

Finally, the composite similarity formula (17) was calculated by weighted assignment of the formula. 17 $S (A, B) = 0.1 S (A_{1}, B_{1}) + 0.1 S (A_{2}, B_{2}) + 0.3 S (A_{3}, B_{3}) + 0.3 S (A_{4}, B_{4}) + 0.2 S (A_{5}, B_{5})$

2.3.2

Design of a method for recommending students’ preferred exercise prescriptions

Through the analysis of the K-means clustering-based exercise prescription recommendation method and the adjustment method based on students’ preferences, the preliminary personalized exercise prescription will be obtained, and then the final personalized exercise prescription will be recommended after the adjustment of the personalized exercise prescription based on students’ preferences. The specific flow of the algorithm is shown in Figure 3.

3

Results and discussion

3.1

Analysis of classification experiment results

A random sample of 6000 male and 6000 female test data from a university’s physical education teaching database was used as experimental data. It was analyzed using K-means algorithm. BMI (weight/height squared), lung capacity, body flexion, standing long jump, 50-meter run, 1000-meter run (male) or 800-meter run (female), and pull-ups (male) or sit-ups (female) were used as inputs for men and women, respectively.

The determination of k value is mainly in the Chengdu Sports Institute of sports experts to ask for advice and discussion, it is recommended that the kinds of 8-15, too many or too few categories of features are not obvious enough, the prescription is not targeted enough to formulate a large number of tests, the calculation of the sum of squares of the error in the cluster for a number of hierarchical clustering, after selecting the optimal center point, the best clustering results, and ultimately selected 10 when observing the various types of specific data classified in all kinds of clustering in the various types of features are most Obviously under, the cluster characterization, exercise prescription formulation.

The center point of K-means algorithm clustering is must be a point in the data with a certain reference value, k is 10, which means that men and women will be divided into ten categories each. The number of iterations is selected 100 times, if the set of center points does not change or reaches the highest number of times has not converged then the last set of center points is selected as the center point. The center point vectors for boys are shown in Table 2.

Table 2.

Boys center point vector

Set id	Center point vector
0	[21.92 5086 234 23 8 9 250]
1	[16.84 4560 226 15.9 7.09 6 250]
2	[19.02 3774 231 15 7.6 11 251]
3	[20.02 4320 236 18 7.4 10 242]
4	[19.17 4068 243 11.7 7.3 15 257]
5	[19.75 6207 231 13 7.35 19 235]
6	[20.13 5375 229 18 7.26 1 253]
7	[20.96 5695 245 22.7 7.65 3 253]
8	[23.35 4865 224 18 7.7 3 260]
9	[21.66 3377 236 13 8.25 3 248]

Boys’ center point corresponding data are shown in Table 3. Each vector data in the table corresponding to the physical test project data scores show analysis, as can be seen from the table, the center point of most of the scores in the various classes of large differences, with a certain degree of variability, the initial exercise prescription development, the center point data as a reference quantity.

Table 3.

The specific data of the male center point vector

Categories	BMI (Kg/M)	Predisposition (Cm)	Fixed Jump (Cm)	Lead Up	50m Run (Second)	1000m (Second)	Lung Capacity (Ml)
0	21.92	23	234	9	8	250	5086
1	16.84	15.9	226	6	7.09	250	4560
2	19.02	15	231	11	7.6	251	3774
3	20.02	18	236	10	7.4	242	4320
4	19.17	11.7	243	15	7.3	257	4068
5	19.75	13	231	19	7.35	235	6207
6	20.13	18	229	1	7.26	253	5375
7	20.96	22.7	245	3	7.65	253	5695
8	23.35	18	224	3	7.7	260	4865
9	21.66	13	236	3	8.25	248	3377

Using the K-means algorithm, the girls’ data were clustered to produce the centroid vectors for each of the last classes of girls, and the centroid vectors for girls are shown in Table 4.

Table 4.

Girl center vector

Set id	Center point vector
0	[25.15 3826 191 34 9.41 41 265]
1	[20.13 2865 172 16.5 8.34 32 240]
2	[20.63 3606 166 15 7.9 33 244]
3	[19.91 3074 176 2 9.5 35 46]
4	[19.72 3440 186 22 8.84 37 249]
5	[24.01 4079 176 18 8.54 36 222]
6	[21.45 3256 163 12 10.2 49 225]
7	[18.46 4617 179 19 10.5 38 239]
8	[18.69 2602 165 12 9.5 47 237]
9	[20.71 2247 179 13 9.8 41 248]

The data corresponding to the center point of the girls are shown in Table 5. The fourth data in the girls’ centroid vector corresponds to sit-ups, and the other items are consistent with the boys.

Table 5.

Girls center corresponding data

Categories	BMI (kg/m)	Predisposition (cm)	Fixed jump (cm)	Sit-ups (one)	50m run (second)	800m (second)	Lung capacity (ml)
0	25.15	34	191	41	9.41	265	3826
1	20.13	16.5	172	32	8.34	240	2865
2	20.63	15	166	33	7.9	244	3606
3	19.91	2	176	35	9.5	46	3074
4	19.72	22	186	37	8.84	249	3440
5	24.01	18	176	36	8.54	222	4079
6	21.45	12	163	49	10.2	225	3256
7	18.46	19	179	38	10.5	239	4617
8	18.69	12	165	47	9.5	237	2602
9	20.71	13	179	41	9.8	248	2247

Taking boys as an example, the boys’ physical test clustering results are shown by PCA (Principal Component Analysis) dimensionality reduction, and the distribution of each class of boys is shown in Fig. 4.PCA enables the new low-dimensional data set to retain as much as possible the results of the original data’s variance, but it is important to note that dimensionality reduction by PCA actually loses some information, by looking at the pca. explained_variance_ ratio worth out [0.99653148, 0.0014521], you can see the two principal components retained, the first principal component can explain 99.64% of the original variance, the second principal component can explain 0.12% of the original variance. This means that about 99.89% of the original information is still retained after being reduced to two dimensions.

Separately analyze the clustering results, carry out result observation and classification result data statistics organization, rank the physical test results, facilitate the exercise prescription formulation as a reference. Take boys as an example, combined with the clustering results, the various categories of boys’ physical test data achievement mean value summing, reference to the center point data for ranking, boys’ physical test data comprehensive ranking shown in Table 6. Where each line represents a category of people, each column of results from 0 to start ranking, 0 indicates that the results of this achievement in the 10 categories of people in the best, 9 indicates that the results of this achievement in the 10 categories of people in the worst. Since BMI size is a range value, more than the range are bad, so the BMI column in the order of size, 0 indicates that the BMI index is the largest, 9 indicates that the BMI index is the smallest. From the table, it can be seen that different categories of the population correspond to the corresponding physical characteristics, such as the male group category 1 combined with the center point data can be seen in the overall BMI is on the low side, the category belongs to the low weight. The pull-up indicator is moderately low, the 1000 meter indicator is at the lower limit of normal values, and the rest of the indicators are in the normal range. The prescription can be formulated by extracting its comprehensive ranking information for relevant formulation.

Table 6.

The boys are ranked in the ranking of data

Categories	BMI From big to small	Lung Capacity	Fixed Jump	Lead Up	Predisposition	50m Run	1000m	Lung Capacity
0	1	7	4	4	9	2	2	0
1	9	4	6	5	2	6	7	9
2	8	6	3	9	8	0	2	8
3	3	5	1	0	1	1	5	3
4	6	8	7	7	5	9	10	6
5	6	0	0	3	6	0	2	6
6	1	3	5	6	2	6	2	1
7	2	2	2	7	10	2	1	2
8	0	4	10	8	9	6	9	0
9	2	6	1	6	0	9	7	2

3.2

Exercise prescription parameter recommendation experiment

Exercise prescription direct recommendation means finding similar groups of current students and then recommending the prescription information with good effect within the group to current students. In this section, 150 students in the test set are randomly selected, and the parameter fusion recommendation method based on similarity fusion computation and the direct recommendation method based on cooperation are utilized to recommend the parameters of exercise prescription for these 150 students respectively, and the prescription effect level under the two recommendation methods is obtained by using the prescription effect simulation program. The experimental results of the similarity fusion computation recommendation and the traditional cooperation-based direct recommendation part are shown in Table 7. The experimental comparison of similar fusion computational recommendation and traditional cooperative-based direct recommendation is shown in Fig. 5 (Fig. a shows the effect ratings of 150 students under different recommendation algorithms, and Fig. b shows the average effect produced by the first i students).

Table 7.

Similar fusion computing recommendation and traditional

Sportsman	1	2	3	4	5	6	7	8	9	10	11	12	13	…	200
Ours	4.3	3.7	4.57	3.8	4.3	4.8	3.6	4.5	2.7	5.8	4.5	4.2	3.8	…	4.8
Direct recommendation	3.6	1.8	3.46	3.74	3.6	5.2	4.1	3.2	1.2	4.7	4.1	4.5	2.7	…	4.9

From the experimental results, it can be seen that the overall effect of using the recommendation approach proposed in this paper is better than the approach based on cooperative direct recommendation in terms of exercise prescription recommendation. More importantly, the stability of the effect of using the recommendation method proposed in this paper is better, which can effectively ensure that the effect of the recommended exercise prescription will not be too bad, while the method based on the direct recommendation of cooperation is susceptible to the influence of the noisy case data, and the stability of the effect is poorer. The reason for the higher stability requirements of the recommended effect in the sports prescription recommendation is that the students in other scenarios can distinguish the advantages and disadvantages and the recommended effect has little impact on the students themselves, but the students in the sports prescription recommendation are generally unable to distinguish the advantages and disadvantages of the sports prescription, and the test cycle of the recommended effect of the prescription is longer, resulting in the recommended effect has a greater impact on the students. Therefore, it is important to ensure the stability of exercise prescription recommendation. The experimental results show that the recommendation method used in this paper has a better overall effect and stability compared to the direct recommendation approach.

3.3

Analysis of Teaching Effectiveness

In order to explore the role of the method proposed in this paper on the improvement of the quality of physical education teaching in colleges and universities, a 16-week teaching experiment was carried out with the students of a university as the empirical research object, setting up the experimental class and the control class, respectively; the experimental class used the method designed in this paper for physical education teaching, while the traditional class used the traditional method for physical education teaching.

3.3.1

Analysis of the effectiveness of teaching specialized physical fitness indicators

After 16 weeks of traditional aerobics teaching in the control group and 16 weeks of mixed aerobics teaching in the experimental group, SPSS26.0 was used to conduct a paired-sample t-test to analyze the changes in the specialized physical fitness of the experimental group and the control group before and after the experiment. Independent samples T-test to analyze whether there is a difference between the aerobics-specific physical fitness of the control group and the experimental group after the experiment, the results of the experimental group and the control group after the experiment and the analysis of the special physical fitness as shown in Table 8 (** indicates that the difference is very significant compared with the pre-experimental p < 0.01, # indicates that the experimental group after the experiment compared with the control group has a significant difference of p < 0.05, ## indicates that the experimental group after the experiment compared with (p < 0.01 difference is highly significant compared to the control group).

Table 8.

Results and analysis of the special body quality

Special body quality index	Group	Pre-test (n= 30) M±SD	Post-test (n= 30) M±SD
Predisposition (cm)	Experimental group	11.42±2.12	19.07±2.13
Predisposition (cm)	Control group	18.54±1.49	19.14±17.03
50 meters (s)	Experimental group	6.07±0.87	9.95±8.24
50 meters (s)	Control group	3.4±0.16	9.7±16.12
50 meters (s)	Experimental group	31.85±4.04	55.16±8.83
50 meters (s)	Control group	28.92±4.83	53.36±6.85
The tablet supports (s)	Experimental group	70.4±10.24	94.2±12.57
The tablet supports (s)	Control group	65.38±21.39	80.08±14.45
At the end of 30 seconds	Experimental group	15.8±3.23	19.65±9.19
At the end of 30 seconds	Control group	14.41±2.74	16.64±10.15

From the table, it can be seen that the p-values of the experimental group in aerobics-specific physical qualities before and after the experiment in seated forward bending (flexibility qualities), one-legged standing (balance qualities), plate support and 30-second two-head rise (core strength qualities) are all less than 0.01, i.e., p<0.01, which is a very significant difference, and the experimental group before and after the experiment 50-meter run (speed qualities) does not have the difference and the experimental group’s mean values of seated forward bending, one-legged standing, plate support and 30-second two-head rise after experiment are all improved than before the experiment. Standing, Plank Support and 30-second Two Heads Up mean values after the experiment were all higher than before the experiment. In the control group, the p-value of one-leg stand (balance quality), plate support and 30-second two-head rise (core strength quality) before and after the experiment is less than 0.01, which is a very significant difference, and the balance quality and core strength quality of the students in the control group after the experiment have been improved significantly compared with that before the experiment. It shows that both teaching methods have certain teaching effects on the improvement of students’ flexibility quality, balance quality and core strength quality before and after the experiment.

After the teaching experiment, the experimental group and the control group carried out independent samples t-test, the p-value of the experimental group and the control group in seated forward bending (flexibility quality), standing with eyes closed on one foot (balance quality), plate support and 30-second two-head rise (core strength quality) were all less than 0.05 i.e., p<0.05, which indicated that there were significant differences between the experimental group and the control group in the quality of flexibility, balance and core strength after the experiment, while the p-value of 50-meter running (speed quality) was greater than 0.05 and did not differ. The p-value of 50-meter run (speed quality) is greater than 0.05, and the mean values of seated forward bending, one-legged standing with eyes closed, plate support and 30-second two-head rise of the experimental group are significantly higher than those of the control group after the experiment, which fully demonstrates that the teaching of the experimental group is conducive to improving the students’ quality of aerobics special qualities in the quality of flexibility, balance, and core strength qualities.

3.3.2

Results and Analysis of Subjects’ Interest in Learning

After the 16-week teaching experiment, the paired-sample t-test was conducted to compare the aerobics learning interest of the experimental group and the control group before and after the experiment, so as to analyze whether there was any statistical difference between the aerobics learning interest of the experimental group and the control group before and after the experiment. Independent samples T-tests were conducted to analyze whether there was a statistical difference between blended and traditional teaching for the students’ interest in learning aerobics after the experiment, and the statistical analysis of the experimental group’s interest in learning after the experiment is shown in Table 9 (** indicates that there is a very significant difference of p < 0.01 before and after the experiment, and * indicates that there is a significant difference of p < 0.05 before and after the experiment. (## indicates a highly significant difference of p < 0.01 between the experimental group and the control group, and # indicates a significant difference of p < 0.05 between the experimental group and the control group).

Table 9.

The statistical analysis of the interest of the experimental group after the experiment

Study interest	group	Before the experiment (n= 30) M±SD	After the experiment (n= 30) M±SD
negativity	Experimental group	19.75±1.08	16.35±3.87
negativity	Control group	22.38±3.1	22.95±1.5
positivity	Experimental group	16.45±1.16	18.46±2.36
positivity	Control group	22±3.82	20.95±4.65
Skill learning	Experimental group	14.17±4.71	23.16±1.2
Skill learning	Control group	17.11±3.93	18.58±1.05
Extracurricular activities	Experimental group	10.44±3.12	22.08±1.32
Extracurricular activities	Control group	12.27±1.07	18.47±3.38
Aerobics attention	Experimental group	12.91±3.31	19.86±3.12
Aerobics attention	Control group	16.35±2.46	17.3±3.46
Total score	Experimental group	73.72±2.49	99.91±2.91
Total score	Control group	90.11±3.13	98.25±4.43

From the table, it can be seen that the mean value of negativity in the experimental group before the experiment is 19.75, and the mean value after the experiment is 16.35. The mean value of positivity before the experiment is 16.45, and the mean value after the experiment is 18.46, and from the change of the mean value, it can be seen that the negativity of the experimental group before and after the experiment is reduced and the positivity is increased for the aerobics students. Through the paired-sample t-test it can be seen that the p-value of all the dimensions in the learning interest of the experimental group before and after the experiment is less than 0.05, which indicates that there is a significant difference between the experimental group before and after the experiment in learning interest. The mean value of negativity in the control group was 22.38 before the experiment and 22.95 after the experiment, and it can be seen from the magnitude of the change in the mean value that negativity also decreased, and the mean value of positivity in the control group was 22 before the experiment and 20.95 after the experiment, and positivity increased. From the table it can be seen that the mean value of the negativity of learning in the experimental group after the experiment is 16.35, and the mean value of the negativity of learning in the control group is 22.95, from the mean value it can be seen that the negativity of the control group after the experiment is higher than that of the experimental group, and the p-value of the negativity of the interest in learning in the control group before and after the experiment and the after-school activities are both less than 0.05 indicating that there is a significant difference in all of them, and the other three dimensions do not have a difference. The independent samples t-test shows that the p-value of learning positivity between the experimental group and the control group after the experiment is 0.012, i.e., the p-value is less than 0.05, indicating that there is a significant difference in learning positivity between the experimental group and the control group after the experiment. The negativity of learning interest of the experimental group was lower than that of the control group after the experiment, and the mean values of the dimensions of skill learning, extracurricular activities and aerobics concern were all significantly higher than those of the control group, and the p-values of 0.000, 0.004, 0.006, 0.001 were all less than 0.01 by the t-test of independent samples, which indicated that there were very significant differences between the experimental group and the control group after the experiment in terms of negativity of aerobics learning, skill learning, extracurricular activities and aerobics concern have very significant differences.

3.3.3

Statistical analysis of overall performance

After 16 weeks of teaching experiments with two different teaching methods, the composition of students’ total performance includes two parts, namely, process evaluation and summative evaluation, i.e., total performance = usual performance (60%) + assessment performance (40%), and the assessment performance includes theoretical assessment performance and technical assessment performance, and through the conversion of each part of the results of the experimental group and the control group of the total performance of aerobics was statistically analyzed, and the total aerobics performance of the grades Statistics are shown in Figure 6. The statistical analysis of the total performance of the experimental group and the control group after the experiment is shown in Table 10 (p < 0.01 The difference is very significant). From the figure, it can be seen that in the total aerobics performance of the experimental group, 13 people were excellent, 16 people were good, and 1 person passed. The number of excellent people in the control group is 7, the number of good people is 8, and the number of passing people is 15. Through the figure, it can be clearly seen that the number of excellent and good people in the experimental group’s total aerobics performance is more than that of the control group, so the experimental group adopts the method proposed in this paper for physical education to be conducive to the improvement of the students’ academic performance. Then Table 10 shows the independent samples t-test for the total achievement of the experimental group and the control group. As can be seen from Table 10, the mean value of the total achievement of the students in the experimental group after the experiment is 91.78±3.14, and the mean value of the total achievement of the students in the control group is 85.76±3.64, and it can be seen that the mean value of the experimental group is significantly higher than that of the control group from the mean value of the total achievement, and it can be seen that the p-value of the total achievement of the experimental group and the control group after the experiment is less than 0.01 through the independent samples T-test, i.e., p<0.01, which indicates that the experimental group and the control group are in a better position to improve students’ academic achievement after the experiment. There is a very significant difference between the experimental group and the control group in terms of total performance.

Table 10.

Statistical analysis of the general results

Project	Experimental group	Control group	t	p
Total score	91.78±3.14	85.76±3.64	4.41	0.000

4

Conclusion

This paper proposes an exercise prescription recommendation algorithm based on K-means clustering algorithm to explore the optimization path of the quality of physical education teaching in colleges and universities. The experimental conclusions of this paper are as follows:

After using the method proposed in this paper for physical education teaching, there are 13 excellent people in the experimental group in the total performance of aerobics, while the number of excellent people in the control group is 7. The number of excellent people in the total performance of aerobics in the experimental group is more than that of the control group, so using the method proposed in this paper for physical education teaching is conducive to the improvement of the students’ physical education performance. In addition, the mean values of negativity in the experimental group before and after the experiment were 19.75 and 16.35, respectively, and the mean values of positivity in the experimental group before and after the experiment were 16.45 and 18.46, respectively, which shows that the negativity of students before and after the experiment in the experimental group was reduced and the positivity was increased.

Langue:: Anglais

Périodicité:: 1 fois par an
Sujets de la revue:: Sciences de la vie, Sciences de la vie, autres, Mathématiques, Mathématiques appliquées, Mathématiques générales, Physique, Physique, autres

RSS Feed de la revue

Categories	BMI From big to small	Lung Capacity	Fixed Jump	Lead Up	Predisposition	50m Run	1000m	Lung Capacity
0	1	7	4	4	9	2	2	0
1	9	4	6	5	2	6	7	9
2	8	6	3	9	8	0	2	8
3	3	5	1	0	1	1	5	3
4	6	8	7	7	5	9	10	6
5	6	0	0	3	6	0	2	6
6	1	3	5	6	2	6	2	1
7	2	2	2	7	10	2	1	2
8	0	4	10	8	9	6	9	0
9	2	6	1	6	0	9	7	2

Categories	BMI From big to small	Lung Capacity	Fixed Jump	Lead Up	Predisposition	50m Run	1000m	Lung Capacity
0	1	7	4	4	9	2	2	0
1	9	4	6	5	2	6	7	9
2	8	6	3	9	8	0	2	8
3	3	5	1	0	1	1	5	3
4	6	8	7	7	5	9	10	6
5	6	0	0	3	6	0	2	6
6	1	3	5	6	2	6	2	1
7	2	2	2	7	10	2	1	2
8	0	4	10	8	9	6	9	0
9	2	6	1	6	0	9	7	2

An Exploration of the Path of Information Technology to Strengthen the Teaching Quality of Physical Education in Colleges and Universities

Yage Yang

Publié en ligne: 23 sept. 2025

Reçu: 14 janv. 2025

Accepté: 19 avr. 2025

DOI: https://doi.org/10.2478/amns-2025-0968

Mots clésK-means clustering, Exercise prescription, Collaborative filtering recommendation, Information technology, College physical education teaching

© 2025 Yage Yang, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Mots clés
K-means clustering, Exercise prescription, Collaborative filtering recommendation, Information technology, College physical education teaching

Categories	BMI From big to small	Lung Capacity	Fixed Jump	Lead Up	Predisposition	50m Run	1000m	Lung Capacity
0	1	7	4	4	9	2	2	0
1	9	4	6	5	2	6	7	9
2	8	6	3	9	8	0	2	8
3	3	5	1	0	1	1	5	3
4	6	8	7	7	5	9	10	6
5	6	0	0	3	6	0	2	6
6	1	3	5	6	2	6	2	1
7	2	2	2	7	10	2	1	2
8	0	4	10	8	9	6	9	0
9	2	6	1	6	0	9	7	2