Page 112 - ASBIRES-2017_Preceedings

Alagalla & Jayakody

Figure 1: Results by registered data

Table 1 displays the results
4 RESULTS according to the number of days activated
As mentioned previously MOOC by the students. Status of the course
data analytics consists of two sections such completion was selected as the class type.
as graph Analysis and data Analysis. Hive
and hbase were used to get more results Following table 1 illustrates the
using querying the data from dataset. accuracy of the classification model by
Accuracy of the output results were taken changing the training data set portion and
by analyzing number of days activated, the classification algorithm.
number of chapters read, number of video
tutorial were used by students. Table 1: Results by number of days
activated
4.1 Graph Analysis

According to Figure 2, it displays
the registered user’s behavior to their 60% training 70% training 80% training
educational level such as bachelors, masters
and PhD. Further, it can be used to compare MultinomialNB
different educational level with number of 0.96 0.96 0.96
days activated. Further, this can be used to BernouliNB 96.67 96.67 96.67
identify the types of people mostly used the SGDClassifier
MOOC’s As a result; administrators can 95.67 95.68 81.45
create another process to motivate them to SVC_classifier 95.68 95.68 95.79
complete the course.
LinearSVC_cla 95.66 95.68 95.82
Figure 3 illustrates the variation of ssifier
registering students on MOOC’s by time MNB_classfier 95.67 95.68 92.16
frame. This graph can be used to get a clear Logistic 95.67 95.67 96.07
idea about the student’s registration time Regression
frames. Table 2 displays the results
4.2 Data Analysis Results according to the number of video tutorials
watched by the student for number of
Data mining techniques were
applied to generate the best classification machine learning algorithm
model.

102