Skip to main content

Table 18 Activity classification and personalized recommendation generation for P-1 in the MOX2-5 dataset

From: AI and semantic ontology for personalized activity eCoaching in healthy lifestyle recommendations: a meta-heuristic approach

Days

The best performing classifier model(s) used in transfer learning

Actual activity level on day-n

Activity level predicted on day-n after incremental learning

Daily achieved score predicted

Propositional variable

Day-1

GB classifier

Low Active

Low Active

1

A-2, A-6, A-8, A-10, A-15

Day-2

Low Active

Low Active

1

A-2, A-6, A-8, A-10, A-15

Day-3

Medium Active

Medium Active

3

A-4, A-6, A-9, A-11, A-14

Day-4

Active

Active

2

A-3, A-6, A-8, A-10, A-15

Day-5

Low Active

Low Active

1

A-2, A-6, A-8, A-10, A-15

Day-6

Low Active

Low Active

1

A-2, A-6, A-8, A-10, A-15

Day-7

Low Active

Low Active

1

A-2, A-6, A-8, A-10, A-15

Weekly Score

Achieved = ∑Daily_achieved_score_predicted = (1 + 1 + 3 + 2 + 1 + 1 + 1) = 10

-

Prediction accuracy

∑Daily_achieved_score_actual—∑Daily_achieved_score_predicted = 0

-

Difference

Defined_goal_score—Achieved = (21 – 10) = 11

-

Weekly

Recommendation

Formal message—“You are 11 points behind to reach your weekly goal. Work hard on the following week.

Informal message – “Improve your performance to meet the goal!

A-17, A-13