Investigation on college students’ consumption ability help classify them as from rich or relative poor family, thus to distinguish the students who are in urgent need for government’s economic support. As canteen consumption is the main part of the expenses of the college students, we proposed the adjusted K-means clustering methods for discrimination of the college students at different economic levels. To improve the discrimination accuracy, a broad learning network architecture was built up for extracting informative features from the students’ canteen consumption records. A fuzzy transformed technique was combined in the network architecture to extend the candidate range for identifying implicit informative variables from the single type of consumption data. Then, the broad learning network model is fully trained. We specially designed to train the network parameters in an iterative tuning mode, in order to find the precise properties that reflect the consumption characteristics. The selected feature variables are further delivered to establish the adjusted K-means clustering model. For the case study, the framework of combining the broad learning network with the adjusted K-means method was applied for the discrimination of the canteen consumption data of the college students in Guangdong province, China. Results show that the most optimal broad learning architecture is structured with 14 hidden nodes, the model training and testing results are appreciating. The results indicated that the framework was feasible to classify the students into different economic levels by analyzing their canteen consumption data, so that we are able to distinguish the students who are in need for financial aid.

Optimal human resources allocation asks to employ a person to work in the position corresponding to his/her ability. Employment competence is the key feedback to the cultivation of college students’ working ability. The data relationship needs to analyze between the in-school cultivation items and the working abilities required by the companies. Machine learning framework is introduced to study the companies’ responses to the cultivation of college students. In this work, a dual-network architecture is built up for statistical modeling evaluation of college graduates’ working ability in consistence with their job position and remuneration. A requirement network and a cultivation network are constructed for extracting features from the original working ability data required by companies and cultivated ever in school. The networks are fully trained by adaptively tuning the linking weights. The extracted features are fused together to estimate the working competence of each target sample/person. To evaluate the dual-network model, a modeling index system is designed, including proposing a total evaluation index calculus for the dual-network model, and a variable importance index from the original data. The samples are consequently ranked by the model predicted index and by the variable importance index, respectively. The ranking difference is used to evaluate the prediction efficiency of the dual-network model. Experimental results show that the dual network architecture is feasible to establish statistical models for the evaluation of college graduates’ in-school cultivated working ability in consistence with the company’s required working ability at their job position and their deserved remuneration.

The quality of graduates is the key factor in evaluating the cultivation effect of colleges and universities. Quantification of whether the graduates qualify for their working post in companies and industries provides conduction for further college cultivation reform enhancement. In this work, we proposed an adaptive multivariate neural network architecture for fusion evaluation of college student cultivation. Specifically, we designed a questionnaire to collect data on the current working status of 1231 graduates and recorded 32 in-school training items categorized into four different modules. For quantitative evaluation, 10 indices of career-require competence were set to describe the graduates’ job abilities. The fused contribution of the in-school training items to the career-required competence was predicted by the multivariate network model with the linking weights adaptively trained. A comprehensive contribution matrix was generated by discrete PCA multivariate transforming to provide a digital reference for the network training. A 7-level scoring system was designed for quantifying the contribution matrix. For model optimization, the network structure was tuned by testing a different number of hidden nodes. The model was trained and optimized to reveal the direct correlation between college cultivation and job-required abilities. Experimental results indicated that the methodology we proposed is feasible to evaluate the cultivation mode in colleges and universities, theoretically and technically providing positive directions for colleges and universities to make their cultivation reforming, as to enhance the quality of their graduates.