Surveying the knowledge of pregnant women towards sport activities during pregnancy using data mining algorithms

The purpose of this study is to research the knowledge of pregnant women towards sport activities using data mining algorithms. Statistical population includes all healthy pregnant women referring to health centers in Gorgan city (Iran) in 2014 from which 429 were chosen as the sample using cluster random sampling. The questionnaire included 65 questions in 6 sections each relating to one knowledge level. Data related to each knowledge level were categorized by decision tree algorithms (CHAID, CART C5.0, QUEST) to predict general knowledge with 3 knowledge descriptions (good, medium, poor) and 5 knowledge descriptions (very good, good, medium, poor, very poor) and then were compared. Also the relationship of these knowledge levels was compared using regression algorithms and SVM. Results show that most of the population has a good and medium knowledge and their knowledge about sport during pregnancy is suitable. In predicting the level of knowledge using decision tree in both prediction level (5 label and 3 label), C5.0 algorithm had the most accurate prediction. Also in comparison, SVM algorithm and SVM regression algorithm had better results with the least error. As a result, it can be said that Extracted rules from algorithms helps to estimating the level of knowledge faster than traditional statically way and provide education regarding exercises during pregnancy for the health of mother and fetus.