运动穿戴设备数据驱动的校园文化需求预测模型

A Data-Driven Campus Cultural Demand Prediction Model for Sports Wearable Devices

摘要:

运动穿戴设备数据构成具有多维属性，直接利用其预测校园文化需求会导致预测结果存在偏差，影响配置文化资源的利用率。开展了运动穿戴设备数据驱动的校园文化需求预测模型研究。在多尺度时域上对原始运动穿戴设备数据特征维度进行分块化处理后，分别结合宏观外生变量和微观外生变量的子序列步长将分块上的数据分解为不同维度的参数，以此独立表征运动穿戴设备数据的多维属性；利用Wasserstein 距离度量不同维度参数与校园文化需求之间的关系，并采用多维映射的方式，在Wasserstein 距离构成的平面上输出分解后运动穿戴设备数据对应的校园文化需求预测结果。测试结果表明，设计模型能够准确预测校园文化需求，配置资源的利用率稳定在50.0%-70.0%区间。

Abstract:

The data composition of sports wearable devices has multidimensional attributes. Directly using them to predict campus cultural needs can lead to biased prediction results and affect the utilization of cultural resources. Therefore, a data-driven model for predicting campus cultural needs using sports wearable devices has been developed. After partitioning the feature dimensions of the original sports wearable device data in the multi-scale time domain, the data on the blocks is decomposed into parameters of different dimensions by combining the sub sequence step sizes of macro exogenous variables and micro exogenous variables, in order to independently characterize the multidimensional properties of sports wearable device data; Using Wasserstein distance to measure the relationship between different dimensional parameters and campus cultural needs, and using multidimensional mapping, output the decomposed sports wearable device data corresponding to the predicted campus cultural needs on the plane formed by Wasserstein distance. The test results show that the designed model can accurately predict the demand for campus culture, and the utilization rate of allocated resources remains stable in the range of 50.0% -70.0%.