Predictive Education

Inspiration

My project is all about helping students succeed. I want to predict which students might face challenges early on, so they can get the support they need in school. The goal is to ensure every student has a chance to excel.

What it does

My project uses machine learning to predict which students might face difficulties in school. By identifying these challenges early, teachers can provide them with the help they need, ensuring a better chance for success in their assignments and overall academic journey.

How we built it

In my project, the provided code serves as the backbone for predicting student test scores based on the hours of study invested. Leveraging TensorFlow and Keras, I've constructed a neural network model aimed at understanding the relationship between study hours and academic performance. The dataset is thoughtfully partitioned into training and testing subsets. The model architecture comprises three dense layers, with the first two layers featuring 256 units each, activated by the sigmoid function. The final layer, utilizing a linear activation function, consists of a single unit, which aligns with the regression task. Training is performed over 500 epochs, utilizing the Adam optimizer to minimize the mean squared error loss. Subsequently, I assess the model's effectiveness by evaluating its performance against the test dataset and generating predictions on how well the student would do compared to the data.

Challenges we ran into

One significant hurdle was sourcing and preparing the data. Finding a comprehensive dataset that included relevant variables and was suitable for training a predictive model required a substantial effort in data collection and cleaning.

Accomplishments that we're proud of

I'm especially proud of how this project aligns with my passion for helping students, including those like me, to succeed academically. Building a predictive model that can provide early insights into potential academic challenges based on study hours feels like a meaningful step towards creating personalized support systems.

What we learned

Through this project, I have learned the practical application of machine learning and neural networks in predicting student test scores based on study hours.

What's next for Predictive Education

What's next for my project is to enrich the predictive model by incorporating more parameters that could impact students' test scores. I plan to expand the dataset with additional features such as attendance records, prior test scores, socioeconomic indicators, and study environment conditions. Furthermore, I aim to fine-tune our neural network by experimenting with various hyperparameters, including the number of layers, units per layer, learning rate, batch size, and regularization techniques like dropout. By incorporating these enhancements, I anticipate the model will become more accurate and versatile in predicting student test scores based on an array of influencing factors, ultimately providing a more comprehensive understanding of student performance.