Over the past two years, the School of Education at The University of Zambia has recorded a high poor academic performance of students in the ICT 1110 course. There could be many reasons as to why the performance rate has drastically dropped and reasons could be no less related to certain academic markers that can foretell student learning outcomes such as class attendance, workload, less course content interactions, unfamiliarity with computer software, demographics, educational backgrounds and educator-student interactions.

This project proposes to create a model that will involve the use of machine learning in order to train and test data from certain input features that will predict students who are at risk of failing ICT 1110. Once the collected data has been pre-processed into datasets from selected data sources such as assessments results, student demographics, student’s academic background, and interactive logs with the Moodle, lecture/tutorial attendance, and instructor feedback. Furthermore, an appropriate machine learning algorithm will use the dataset to produce a trained classifier model. This trained classifier will be able to take in new data rows and be able to predict the label of that new data. Educators, who could be lecturers or tutors, would be able to log into the web model to identify the at-risk students and the students who are thus affected will receive notification once there are online so that they are informed that they are at risk of failing the ICT 1110 course.

The project will use a mixed approach and will be built around the cross industry standard process for data mining (crisp dm). It involves six phases.

The first phase is business understanding. Which involves general and specific objectives, situational analysis and broad goals of data mining process.

The second phase is data understanding which involves identification of data sources , initial data collection, data quality assessment.

The third phase is data preparation. which is related to data selection process, data pre-processing , data transformation and derivation, data merging, data formatting and also data description.

The fourth phase is modeling. defines the model components, how it behaves and how to interpret. It also involves evaluating the various alternative techniques that can be integrated with the model

The fifth phase is evaluation.which involves evaluation techniques to be used considering the efficiency vs effectiveness of the algorithms.

The sixth phase is deployment. Which involves to determine the model results will be presented to the end users that will need to use the model results.

The anticipated outcome of this project is that the model will be able to predict and classify the students who will be at risk of failing. The benefits which the software is going to bring is that, students will be able to know in time as to whether they are at risk of failing hence they will be able work extra hard and possibly avoid the incident from happening. The lecturers will also be able to know which student is at risk of failing a course, hence taking measures on how best such student can be helped.