Introduction

This projects seeks to understand and effectively visualize important predictors of student depression. The unfiltered dataset includes 18 variables and 27894 observations across cities in India. The response variable, Depression, is binary: either yes the student has depression or no they do not. Following an initial exploration of the data, the explanatory variables of financial stress (categorical), sleep duration (categorical), and age (numerical) were found to be important to predicting depression incidence. Source of Data: https://www.kaggle.com/datasets/adilshamim8/student-depression-dataset

Figure 1

Heatmap of Depression Incidence by Sleep Duration and Financial Stress Levels.

Figure 1 shows that, across the whole dataset, lower levels of sleep and higher levels of financial stress are correlated with higher depression incidence rates. The range of depression incidence is quite striking – only 24.2% of students are depressed who get >8 hours of sleep and have very low financial stress, compared to 86.6% of students who get <5 hours of sleep and have very high financial stress.

Figure 2

Barplot of Sleep Duration and Financial Stress Categories Across Three Age Classes

Figure 2 shows that the trend of financial stress and sleep duration highlighted in Figure 1 is consistent across age classes. Interestingly, however, for those of the same financial stress and sleep duration class, as you age increases average depression prevalence decreases. So far, it appears younger age, higher financial stress, and less sleep are correlated with higher depression rates.

Figure 3

Interactive Plot of Student Depression Rate and Predictors Across 30 Cities in India

Figure 3 shows how the trends revealed in Figure 1 & 2 vary across different cities. Cities with the lowest student depression rates are close to 50% on average. Cities with the highest student depression rates are close to 70% on average.

Both of these numbers are too high!! Here’s to hoping we find ways to reduce student depression prevalence. (Get some sleep!)