FORECASTING METHODS AND APPLICATIONS
Number of Marriages in Turkey
May 17th, 2013
1. Introduction and Problem Definition
Marriage is an important concept for Turkey both socially and economically. The reason why we choose the number of marriages is to analyze the impact of changing and developing deterministic factors on the number of marriages in Turkey. These factors that we choose to analyze the effect on marriage are golden prices, unemployment rate, temperature and CPI. Golden prices are important since the weddings in Turkey include receiving golden as a gift for the couple. If the golden prices are high, people tend to get married at these times more than any other time that golden prices are low. Unemployment rate presents the economic situation of Turkey therefore economic situation of the couple. Another independent variable is CPI which represents the consumer price index. Finally, temperature is the most important independent variable in our opinion, since people tend to get married in summer season mostly. Our time interval is the years between 2005 and 2011. The data is monthly gathered, so we have 12 observations for each year. Therefore, we have 84 observations in total. There are some similar studies in literature like researching the impact of some factors on the duration of marriage, done by Duygu Özşen as her master’s thesis called “Analysis of factors that effects marriage duration with cox regression” . She made a research about factors that effects marriage duration and focuses on appropriate partner choice, status before marriage, number of children, age, occupation, education level, income level, gender, etc. She made an analysis with Cox regression to analyze the impact of the factors. 2. Methodology
First of all, in order to estimate better results and make better interpretations, the data is collected from TUİK website. According to the gathered data; regression, smoothing and Winter’s method will be used in order to analyze and interpret the relevant data sets. Then, the impacts of deterministic factors on the marriages in Turkey would be figured out. In order to make reliable estimations and then interpret them correctly, Minitab is used. Minitab helps us to get many graphs and estimation according to the data sets and their impacts with number of marriages through years 2005 – 2011. 3. Description of Data Sets and Related Variables
Our dependent variable is number of marriages in Turkey. As it is mentioned before, there are four independent variables: golden prices, CPI, temperature and unemployment rate. Our data sets are in Appendix 1 and we have 84 different observations (monthly data for seven years). Dependent variable
Y: number of marriages
X2: Golden prices
X4: Unemployment rate
According to time plots of variables (see Appendix 2), it’s observed that number of marriages and temperature shows seasonality through the seven years. However, golden prices and CPI show trend. Unemployment rate does not show any seasonality or trend.
4. Analysis - Description of the Time Series Methods used After methodology and description of data sets and related variables, first we need to focus on the behavior of data by considering each dependent and independent variable separately. Thus, as it is seen in Appendix 3 we graphed autocorrelations of each variable by Minitab. To strengthen our analysis we also graphed partial autocorrelations of same variables in Appendix 4. In addition, cross correlations of variables can be seen in Appendix 5. First step is completed and we need to decompose the given data of number of marriages per month. Our linear trend equations is; Y = 53599 + 49, 69*t
Then, components of decomposed data are analyzed. Detrended data shows that number of marriages have a strong seasonality component. Also seasonally adjusted...