The Student Net- Using Machine Learning Algorithms to Address our Failing Guidance System
By Francesca Bizzarri
The guidance system is failing its students.
Lily, whose real name has been kept anonymous, is a school counselor from Newfoundland. Her services are used at three different secondary schools; this means that she is assisting around 500 different students at a time. She constantly wishes that she could be in three places at once and truly wants to support all her students, but it simply is not possible with the sheer amount of students she has to service.
On the extreme end of Ontario secondary schools, the ratio of guidance counselors to students is 1:826. That means just one counselor is trying to provide support to almost 850 students!!
The core task of school counselors is to support student’s career development, supporting personal, social, and educational development. However, assistance for university planning and career planning often takes a back seat to urgent mental health support in times of crisis and consequently, many students feel like their guidance counselors aren’t prepared or able to help them with career and university planning.
This insane number of students that counselors are assigned to assist greatly impacts a counselor’s ability to perform effectively and many students end up receiving insufficient support to help prepare them for entering the real world. As a group of students, we have seen this firsthand. Oftentimes, we find that our counselors are not able to provide us with quality career planning support and we find ourselves having to remind our counselors to actually help us.
We decided to put out a survey about the current guidance system for our fellow TKS students to investigate the problem firsthand. Here are a few results from our survey.
68.6% of students said that they haven’t had a meeting yet with their guidance counselor. and it’s the middle of October! 33.3% of students indicated that they have received no communication from guidance and 31.4% of students said that they themselves had to reach out to guidance for help.
The current system that we have in place simply assumes that counselors can effectively juggle all of these administrative duties in addition to effectively assisting hundreds of students in planning for their futures- but the problem doesn’t stop there. The group that suffers the most due to the current system are the ones who are in most need of guidance counselors; at-risk students. In fact, the graduation rates in Ontario range from 68% to 77%. This means that around 23% of Ontario youth were unable to complete their secondary education. According to Pathways Education, the drop out rate can increase to as high as 50% or more, in low-income communities.
So, what factors contribute to a student being at-risk for dropping out?
The main factors that contribute to a student having difficulty graduating or dropping out entirely are school environment, personal factors, family influences, health, and community environment. In order to determine whether or not a student is at risk, all of these factors must be considered.
PROBLEM BREAKDOWN
This system of overwhelmed and unorganized guidance resources is failing to identify at-risk students and provide them with sufficient career and university planning services, allowing them to fall through the cracks.
Why is it so difficult for the guidance system to identify and support these at-risk students?
- Using the grading system alone to identify struggling students is ineffective.
Traditionally, schools have solely relied on using metrics like GPAs absence rates, tardiness, and feedback from teachers to identify at-risk students. Not only are human judgment metrics like these static and error-prone, but they are also limited which inhibits the ability of these metrics to accurately identify at-risk students in a variety of different environments.
2. Schools don’t have adequate resources of determining if a student is at risk of dropping out of school or graduating with difficulty.
As mentioned above, guidance resources are stretched very thin and allow at-risk students to slip through the cracks. Currently, all of the weight for supporting these students is placed on the counselors. This weight needs to be redistributed and the limited guidance resource has to be directed properly to make this system efficient.
Why is addressing this problem significant?
When guidance counselors don’t provide support to students who need it, they end up being set up to fail and have a much harder time entering the workforce or even graduating. We want and need students to graduate and succeed so that they can actively contribute to society instead of being a weight. High school dropouts face extremly limited future oppritunites. With fewer skills, they are less likely to find a job, earn a living wage, and more susceptible to illness and crime. Dropping out of high school not only affects the individual, but also those who surround the individual. As the number of highschool dropouts increases, their impact on the economy and society as a whole increases as well. In fact, high school dropouts cost Canadian taxpayers around $1.3 billion in social services and criminal justice expenses each year.
INTRODUCING OUR SOLUTION- THE STUDENT NET
What are we doing about it?
The most obvious solution that comes to mind with this problem is to just hire more guidance counselors to lessen the load. But with limited funding and administrative hurdles, this solution has not been efficiently implemented. This is where our service, The Student Net comes in.
To address the failure of our guidance system in supporting at-risk students, we aim to develop machine learning algorithms to analyze a multitude of factors to identify at-risk students. Our program will be able to identify students who are at-risk for dropping out or having difficulty graduating and direct the guidance system to move their resources to support this student consequently making the system much more efficient and reliable.
The platform- How does it work?
Step 1️⃣: School Districts or cities apply for this service through a secured website.
Step 2️⃣: We collect the data from the specific schools and communities to make our recommendations.
Step 3️⃣: Our algorithm takes the data collected and identifies the students at risk for dropping out or difficulty graduating and indicates how at risk they are using a 1–10 urgency scale.
Step 4️⃣: This information is shown to the particular school through the secure account-based website.
Step 5️⃣: Guidance counselors can then access more information about why our algorithm made these predictions and look at what specific aspects contribute to the student being at-risk.
Our service would address the need for a much better organization of guidance services and the inability of the system to support at-risk students. Our aim is to identify at-risk students and then direct guidance resources to the students who need support before serious consequences, such as dropping out, even happen.
How are we going to avoid bias?
One of the hurdles with this service is the potential for bias. Our algorithm looks at data and trends to make these predictions on whether or not a student is at risk. Bias in data sources and bias within the algorithm has the potential to wrongfully identify a student as at-risk. For example, if a large number of students from one neighborhood are identified as at risk, the algorithm could potentially label all the students from this neighborhood as at-risk.
In order to mitigate this potential for bias, we would first and foremost gather data from diverse sources and also include diversity in our team as the first people to notice issues with biases are most commonly minorities. We would also implement the use of a debiasing tool through the AI fairness project. This system would work by running a range of metrics against a class label that quantifies the model’s bias toward particular members of the class. The AI Fairness 360 library has 10 debiasing approaches that could be applied to our model.
Furthermore, we would implement a system where before taking any serious action or alerting family, counselors simply meet with the student to get to know them and assure that our predictions were valid. The guidance counselors would then be able to tell us if our predictions were correct through a feedback system and we would then adjust the algorithm accordingly.
How do we know that our solution will actually have an impact on this problem?
We can see the need and impact of our solution through the story of Tommy Paley, a former high school math teacher. A student, anonymously called George in Tommy’s story, was in his math class a year before he committed suicide at the age of 16. George always came in with a bit of a smile on his face and even though he didn’t like math, he still always came. The next year Tommy barely saw George except for a passing hello in the hallway or wave. But one day, in an emergency staff meeting the principal announced that George had jumped in front of a train the previous day. In the note left to his parents, he said he was tired of being bullied about being gay.
Perhaps if George’s counselors were aware of what was going on with the bullying, George’s story would have played out a little differently. If there was something that could have directed guidance resources to George before he made the decision to step in front of that train, he would still be here today. This is the gap that our service aims to fill.
Future Goals For this project
Once we start to see patterns in data of at-risk students in specific schools or regions, we will know if there is a larger problem to address. Since this project works out the factors contributing to students being at risk, we can use this as a stepping stone to reform education by addressing larger problems like:
- Income inequality
- Large class size impacts
- Lack of specialized teachers
TL;DR
- Guidance counselors are overwhelmed and can’t identify or provide proper support to at-risk students, allowing them to slip through the cracks.
- We developed The Student Net to correctly identify at-risk students and help direct and organize guidance resources so that these at-risk students get the support they need before serious consequences occur.
- We aim to use this platform in the future to address issues like income inequality and education inequality as we collect vast amounts of data and can analyze trends.
Our aim is to implement this algorithm as a provincial board initiative and to eventually directly provide mentors, personal recommendations to struggling students through this platform.
As a group of students ourselves, we can see the flaws within our guidance system because we experience them every day. Understanding that funding for more guidance counselors is difficult to promote, we designed this solution as another way to help our fellow peers to receive the quality support that they deserve.
Thank you for taking the time to learn about our solution to the failing guidance system. To learn more about it, check out the links below!!
Check out our one-pager here
Link to our website is here :)
