Introduction

SkyApp is an educational mobile application developed by Dr. Lucas Hui and his research team to allow easy creation of learning activities by teachers on iPad. SkyApp overcomes the limitations set for paper assessments because data beyond answers and scoring. SkyApp takes advantage of technology to collect data at a fine-grained scale, and thus making SkyApp unique to other educational mobile applications. Examples of fine-grained data in SkyApp includes finger movements, time spent on each question, number of trials and answer of each trial. The analyzing tool to be built in this project will be plugged into SkyApp to support the needs of teachers in understanding students' learning performances and challenges faced.

Objectives

The project aims to provide meaningful information to students and and teachers based on data collected by SkyApp. The information shall be able to support teachers in understanding their students' learning patterns and performances. The teachers will hence be able to deliver students' need.

Analysis reports are to be updated and accessible to teachers on SkyApp. The report is to provide teachers statistical information on students' learning performances throughout students' participation e-learning activities of the class. Also, through data analysis, this project aims propose classification of students' based on their learning patterns. Classification framework to identify students' learning patterns is expected to help teachers in improving their teaching.

Scope & Deliverables

Parameters adopted for data analysis in this project are subjected to database design of SkyApp version 1.22.

First Semester	Second Semester
1) Basic statistics report	1) Detailed statistic report
2) Data processing program	2) Analyzing tool for detailed statistics
3) Draft on classification framework

Approach & Methodologies

Once the app is launched for the pilot test in October, actual data in JSON format and will further process it to be stored in the MySQL database. The actual data contains a large amount of data and requires data cleaning to extract meaningful information for analysis. Python will be used in this step since it is efficient in processing data with different format matters.

R programming is an open-source analytical tool and has advanced graphical capabilities when comparing to Python and SAS, so it will be used to produce analysis reports and diagrams. Data analytic techniques such as clustering, decision tree, and association rule will be applied to find out correlations between the data and students' performances. Afterwards, we will develop a model that can map some combinations of activities patterns to different categories of students.

The first iteration of the research will start in October. In this iteration, SkyApp will start collecting fine-grained data of students. Before the first iteration is completed, sample data will be collection. Sample data will be collected by distributing three exercises and one test to each of the participant. Ideally, forty participants will be contributing data through answering the exercises and tests. Sample data will facilitate our job in building data processing program and basic statistics report before actual data is obtained.

Upon obtaining the actual data, data mining will be carried out. A few hypotheses of students' learning patterns are developed, which are the possible approaches in conducting data mining. For example, the time students spend on questions may be related to their level of concentration. The correctness and number of trials may be a significant track of their learning process. The use of negative emoji may show their lack of interest or confidence, etc.

For this project, typical data mining lifecycle is adopted to ensure the feasibility and efficiency of the project. The stages of lifecycle are discovery, data preparation, model planning, model building, presentation and deployment. Details of each stage can be retrived from the Project Plan document.