How many deaths on a campus are acceptable?

The answer is ZERO.

Using predictive analytics for two main goals:

1) Find current changes in student behavior.

2) Predict future behavior with AI and advanced forecasting techniques.

1. Current changes in behavior:

As a student goes through drastic changes in behavior, we are able to see these affects though different avenues of their life. Example: grades dropping, missing appointments, disciplinary action, etc. If each one of these data points where plotted on a graph, we may see something similar to a falling stock or downtrend. To find the changes in behavior we use a tool such as ARMA (Auto-Regressive Moving Average) to determine if they are on a down trend, and how far they may have shifted from the average behavior. Applying this technique to each data point autonomously will allow us to receive real-time alerts if a student is going through a drastic behavior change.

2. Predicting future behavior changes:

To predict future trends we are relying on Artificial Intelligence concepts and utilizing a state of the art algorithm called LSTM (Long-Short Term Memory). This algorithm allows us to gather student data and learn advanced patterns that a human wouldn’t be able to detect. These patterns also give us the ability to see future insights into the evolution of the students behavior over time, including future predictions.

LSTM’s are used in advanced cases such as voice recognition, language processing, financial and economical analysis.

 

Utilizing both of these methods, we can generate a picture to determine if a student is in need of help or going though a change. This gives us time to help and assist before it’s too late.