Saturday, September 15, 2012

Day 9 - Team E - Ashok Sasidharan

 Today we first analyzed people's preferences towards various types of music using the 'GSS93 subset' file. We used custom table to analyze the music preference first among people with different marital status and then among people belonging to different age groups. The preference towards 4 types of music - Traditional,Soft,Country and Rock - was analyzed. The following 2 tables were obtained

Traditional Soft Country Rock
Mean Mean Mean Mean
Marital Status married -0.043 0.072 -0.076 0.138
widowed -0.247 0.252 -0.086 0.452
divorced 0.114 -0.119 -0.089 0.012
separated 0.565 -0.362 0.465 -0.460
NeverMarried 0.093 -0.213 0.305 -0.637
Traditional Soft Country Rock
Age Categories Mean Mean Mean Mean
18-29 0.326 -0.040 0.345 -0.794
30-39 0.178 -0.183 0.089 -0.201
40-49 -0.002 -0.048 -0.081 0.078
50+ -0.258 0.165 -0.161 0.435

                              



  Permap input files were created for both the cases and they were opened in Permap and active vectors were checked. The following 2 diagrams were obtained.

This showed that more number of divorced people were attracted towards rock music while more people who are not married have affiinity towards country music and less affinity towards rock music. More widowed people preferred soft music while very less separated people preferred soft music. 



 
        It shows that less number of young people(18-29 years) liked Rock music while more number of people in the Upper middle age range (40-49) liked Rock music but less people in that age category preferred country music. Less number of old (50+) people preferred Traditional music and less number of people in the 30-39 age range preferred soft music.

Discriminant Analysis
  We also learned about Discriminant Analysis. It is used to distuinguish distinct sets of observations and allocate new observations to previously defined groups. This analysis was performed on 'bankloan.sav' file. The technique used in discriminant analysis is regression. All variables should be scale and not nominal or ordinal. There are dependent and independent variables. After performing this analysis, a table was generated which gave information on demographics,income details etc of people who have defaulted on loan as well as those who are not defaulters. Its advantages are

  • It generates helpful plots, especially a territorial map, to aid analysis.
  • It still offers the opportunity for classifying cases that are "ungrouped" on the dependent variable.




















No comments:

Post a Comment