你幸福吗？《2010年中国城市居民幸福感调查》可靠性论证！ [推广有奖]

121楼

piaoyao55555 发表于 2011-3-1 20:41:51

真的没有太多让人幸福的事

122楼

crossyeer 发表于 2011-3-1 21:49:08

楼主哦，这个调查面会不会有些狭窄，集中的都是大家这些关注经济类比较多的人，其中学生应该更是大多数吧，那么结果应该有不小偏差吧~题目也许得改改，呵呵

123楼

regressi0n 发表于 2011-3-1 22:58:20

谈到幸福，有愿意探讨幸福函数的坛友吗？下边是我归纳的一点点前人关于幸福函数的认识。先看看这些，也许能节约大家翻阅大量文献的时间。

min∫0+∞[B-U(x)+V(l)]dt
s.t. dk/dt=f(l,c)-x
where,
0 lower limit of integration
+∞ upper limit of integration
B maximum value of satisfaction of demand, conditioned on long-run technical level available to make use of measurable resources or endowment
U(x) utility achieved by spend monetary earning at amount of x
V(l) payment earned by means of providing labour services at amount of l by individual or group of people, and paid by employer or business activity
t time
k capital
x expenditure for consumption

Replacing t of target function with c stated in constraint condition, we have the target function updated as
min∫0+∞((B-U(x)+V(l))/f(l,c))dc
Both expenditure x and labour input l are function of consumption c, so that
f(l,c)–x=(B–U(x)–V(l))/u(x), where u(x): marginal utiltiy
by means of minimizing integrated function, or making partial derivative of it equal to zero in this case; and so that it is clear to see that the left part of this function works as saving ratio (or investment ration), which is vitally important in terms of development, even progress, and the right one as another ratio of difference, between maximum value of satisfaction of demand and sum of utility at cost of x and salary at cost of l, and marginal utility u(x).

124楼

regressi0n 发表于 2011-3-1 23:09:53

幸福(H)作为因变量，H(xi) (i=1,2,...,n)，能罗列并说明这n个x吗？如果能做到这一点，那么计算幸福指数就非常简单了，来个神经网络，什么都万事大吉了。请看：
logit^(-1)H(xi)=(1+exp(H(xi)))^(-1)
无比简单，是不是？只要把H(xi)解出来就是了。想了解这方面算法的坛友，建议看看2000年的诺贝尔经济学奖，好像是2000年的，我有些记不请出了，抱歉。
25# 海洋小公主

125楼

regressi0n 发表于 2011-3-1 23:14:32

好啊，那就请您把眼界放大到随便什么程度都行，罗列一个n列的包含有1000000个observations的表格，好吗？其实多少都行，没有必要来那么个n乘1000000的矩阵。只不过太少了，精度会出问题的。

幸福(H)作为因变量，H(xi) (i=1,2,...,n)，能罗列并说明这n个解释变量x吗？如果能做到这一点，那么计算幸福指数就非常简单了，来个神经网络，什么都万事大吉了。请看：
logit^(-1)H(xi)=(1+exp(H(xi)))^(-1)
无比简单，是不是？只要把H(xi)解出来就是了。想了解这方面算法的坛友，建议看看2000年的诺贝尔经济学奖，好像是2000年的，我有些记不请出了，抱歉。

122# crossyeer