Why the fifth model's amount of parameter is 120? By knowledge from CMP, I know first model is constants+ coffecient of one variable=2. I know second model is 2+(10-1)=11. I know third model is 2+(10-1)+(2-1)(10-1)=20. I know fourth model is 2+(10-1)+(2-1)(10-1)+(6-1)=25. But I have no idea about fifth model.I guess it is 2*10*6,but I don't know the logic. It will resulted in can't count amount of parameter of fourth model, if I just calculate parameters. SA=alpha+beta=2 SA+PT=alpha_i+beta=11 SA*PT=alpha_i+beta_i=10+10=20 SA*PT+NB=alpha_i+beta_i+B_j=10+10+6=26
Hi Ykai The question gives you the 120 SA*PT+NB=alpha_i+beta_i+B_j=2+(10-1)+(2-1)(10-1)+(6-1)=25 SA*PT*NB=alpha_ij+beta_ij=2+(10-1)+(2-1)(10-1)+(6-1)+(2-1)(6-1)+(10-1)(6-1)+(2-1)(10-1)(6-1)=120 you add on all the interactions SA*NB+PT*NB+SA*PT*NB to the previous model Rule for interactions are (n-1)*(m-1) where n and m are the number of parameters you would have used for the effects on their own Thanks Andrea
Thank you for answer.I have understand the method of calculation,but I still have question about display all possible parameter directly. The 25 one,SA*PT+NB, I guess it have 10 kinds of alpha from constant*T_i +10 kinds of beta from T_i*coffecient of x+6 kinds of B_j, so I think it is a total of 26 parameters not 25. Paremeters are alpha_1,alpha_2,alpha_3,alpha_4,alpha_5,alpha_6,alpha_7,alpha_8,alpha_9,alpha_10 beta_1,beta_2,beta_3,beta_4,beta_5,beta_6,beta_7,beta_8,beta_9,beta_10 B_1,B_2,B_3,B_4,B_5,B_6 After I read CH13-section3.3-p25, I think what I thought may be right. Which part of what I said is wrong?
alpha_1 and B_1 would combine, as it is the sum of two constants, working from model 3, so the base level is one bedroom and property type 1, you would then add on a value for the other 5 possible bedroom combinations, so lose B_1 from your list, giving 25 parameters in total OR lose alpha_1