An exploration of the relationship between BMI,eyesight and mid-term test scores

Data

The data is based on data collected from CHINA EDUCATION PANEL SURVEY at CEPS.

Method

Data Cleaning

Delete student samples missing or incorrect essential variables.

Variable Selection

• Independent Variables:
  Body Mass Index calculated by height and weight;
  Whether the student is short-sighted.
• Control Variables:
  Gender,Cognitive test scores, the type of household registration account, whether the student is the only child in the family, the financial condition of the family, the educational expectation of parents and child, the educational degree of mother and father
  
• Dependent Variables:
  Mid-term Chinese, English, Math score.

Linear Regression using sklearn

• Use Linear Regression model to predict the relationship between independent variables, control variables and dependent variables.
• Group student samples based on BMI to explore test performance

Conclusion

For those with low BMI, the higher their BMI, they better their scores.
For those with low BMI or normal BMI, the nearsighted students had better scores. For those with high BMI or overweight, the higher their BMI, the worse their scores.