Dataset.csv

The purpose of this dataset was to use machine learning to predict the opportunity of divorce among single-parent families in Thailand. The use of Krejcie and Morgan tables for sample size calculation considers the proportion of traits of interest in the population, the desired level of tolerance, and the confidence level. The sample size of 52 out of a population of 60 indicates a high response rate, which is a positive aspect of the study. 60 questions in the online questionnaire, only 12 were related to the likelihood of divorce as; <br> <strong>Questions List</strong> 12. At what age did you first engage in sexual intercourse? 17. Did your parent divorce or separate in your childhood? 22. Which were the reasons for deciding to choose a spouse? 24. At what age did you start living together as a couple for the first time? 25. How many life partners did you have? 26. How long have you lived as a couple with your spouses? 27. How often did your arguments with your spouse? 28. What were the reasons for your arguments with your couple? 29. How long did it take from the beginning of the problem until deciding to separate? 30. Which were the reasons for the decision to separate? 31. How many children do you have? 32. How old were you when you had your first child? <br> All questions and answers are typically converted to numeric code to enable machine learning algorithms to understand and analyze the data. For example, question 12 asks How old were you when you had your first sexual intercourse? converts to Q12. The question has the following options: Never had sex, under 20 years old, 20 - 25 years old, 26 - 30 years old, and 31 years old and over are converted to 1, 2, 3, 4, 5, etc. <br> Transforming questions with multiple answers into attributes is a technique in machine learning. This process is called one-hot encoding and involves creating a new binary attribute. For example, question 22 asks What were the reasons for deciding to choose a spouse? The answers are Appearance, Love, Financial, Character, and Family approval will be transformed into attributes Q22-A1, Q22-A2, Q22-A3, Q22-A4, and Q22-A5. If the subject chooses any answer, the value will be substituted for 1 and 0 when not selected. <br> This process prepares the data in a format that is ready for analysis. As a result, all 37 topics have a feature value. Each row represents one spouse's response per couple. The columns represent questions in order of attributes Q12, Q17, Q22-A1, Q22-A2, Q22-A3, Q22-A4, Q22-A5, Q24, Q25, Q26, Q27, Q28-A01, Q28-A02, Q28-A03, Q28-A04, Q28-A05, Q28-A06, Q28-A07, Q28-A08, Q28-A09, Q28-A10, Q28-A11, Q28-A12, Q28-A13, Q29, Q30-A01, Q30-A02, Q30-A03, Q30-A04, Q30-A05, Q30-A06, Q30-A07, Q30-A08, Q30-A09, Q30-A10, Q31, Q32, and Divorce. <br> The attribute Divorce is a class for labeling the data. Representing divorce or marriage by substituting those whose status is married = 0, divorced = 1.