tの統計値を用いたインターセプトと回帰係数の平均

Question

私は変数の2つの行列を持っています：lagcolmean（独立）とDropcolmax（依存）。

head(lagcolmean) 
     MSFT  AAPL  GOOGL 
2 -0.2130120 0.8115149 -0.2093960 
3 -0.1824693 -0.3352958 -0.1845572 
4 0.2175819 -0.6885582 0.9421018 
head(Dropcolmax) 
     MSFT  AAPL GOOGL 
1 0.3837303 3.871109 6.618858 
2 0.6723379 4.369627 7.070757 
3 1.1848480 2.565165 2.079593 

Iは、これら二つの変数を回帰するために、この式を使用する：

fit <- lapply(names(Dropcolmax), function(x){ 
    lm(lagcolmean[[x]] ~ Dropcolmax[[x]]) 
}) 

は、したがって、私は、任意のt値なく3つの傍受及び3つの係数を得ます。今私は、これらの3つの切片とそれに対応する平均値を持つ係数を平均して報告したいと思います。

Answer 1

私はあなたが得ることができるさまざまな出力を表示するためにステップバイステップで機能を変更しています。私の意見では、broomパッケージは、大きなデータフレームに組み合わされた各モデルの有用な情報を提供するので、あなたの人生を楽にしてくれます。次に、適切な列とフィルタを使用して、必要なものをすべて計算できます。

library(broom) 

lagcolmean = read.table(text=" 
MSFT  AAPL  GOOGL 
-0.2130120 0.8115149 -0.2093960 
-0.1824693 -0.3352958 -0.1845572 
0.2175819 -0.6885582 0.9421018", sep="", header=T) 

Dropcolmax = read.table(text=" 
     MSFT  AAPL GOOGL 
0.3837303 3.871109 6.618858 
0.6723379 4.369627 7.070757 
1.1848480 2.565165 2.079593", sep="", header=T) 


# returns list of models with minimal info 
fit1 <- lapply(names(Dropcolmax), function(x){ 
    lm(lagcolmean[[x]] ~ Dropcolmax[[x]]) }) 

fit1 

# [[1]] 
# 
# Call: 
# lm(formula = lagcolmean[[x]] ~ Dropcolmax[[x]]) 
# 
# Coefficients: 
# (Intercept) Dropcolmax[[x]] 
# -0.4819   0.5657 
# 
# 
# [[2]] 
# 
# Call: 
# lm(formula = lagcolmean[[x]] ~ Dropcolmax[[x]]) 
# 
# Coefficients: 
# (Intercept) Dropcolmax[[x]] 
# -1.4706   0.3886 
# 
# 
# [[3]] 
# 
# Call: 
# lm(formula = lagcolmean[[x]] ~ Dropcolmax[[x]]) 
# 
# Coefficients: 
# (Intercept) Dropcolmax[[x]] 
# 1.4289   -0.2371 


# returns a list of models with useful info 
fit2 <- lapply(names(Dropcolmax), function(x){ 
    summary(lm(lagcolmean[[x]] ~ Dropcolmax[[x]])) }) 

fit2 

# [[1]] 
# 
# Call: 
# lm(formula = lagcolmean[[x]] ~ Dropcolmax[[x]]) 
# 
# Residuals: 
# 1  2  3 
# 0.05179 -0.08095 0.02916 
# 
# Coefficients: 
#     Estimate Std. Error t value Pr(>|t|) 
# (Intercept)  -0.4819  0.1430 -3.370 0.184 
# Dropcolmax[[x]] 0.5657  0.1750 3.233 0.191 
# 
# Residual standard error: 0.1004 on 1 degrees of freedom 
# Multiple R-squared: 0.9127, Adjusted R-squared: 0.8253 
# F-statistic: 10.45 on 1 and 1 DF, p-value: 0.191 
# 
# 
# [[2]] 
# 
# Call: 
# lm(formula = lagcolmean[[x]] ~ Dropcolmax[[x]]) 
# 
# Residuals: 
# 1  2  3 
# 0.7777 -0.5628 -0.2149 
# 
# Coefficients: 
#     Estimate Std. Error t value Pr(>|t|) 
# (Intercept)  -1.4706  2.7482 -0.535 0.687 
# Dropcolmax[[x]] 0.3886  0.7465 0.521 0.694 
# 
# Residual standard error: 0.9838 on 1 degrees of freedom 
# Multiple R-squared: 0.2132, Adjusted R-squared: -0.5735 
# F-statistic: 0.271 on 1 and 1 DF, p-value: 0.6944 
# 
# 
# [[3]] 
# 
# Call: 
# lm(formula = lagcolmean[[x]] ~ Dropcolmax[[x]]) 
# 
# Residuals: 
# 1   2   3 
# -0.069114 0.062857 0.006258 
# 
# Coefficients: 
#     Estimate Std. Error t value Pr(>|t|) 
# (Intercept)  1.42885 0.13717 10.417 0.0609 . 
# Dropcolmax[[x]] -0.23707 0.02398 -9.884 0.0642 . 
# --- 
# Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 
# 
# Residual standard error: 0.09363 on 1 degrees of freedom 
# Multiple R-squared: 0.9899, Adjusted R-squared: 0.9797 
# F-statistic: 97.7 on 1 and 1 DF, p-value: 0.06419 


# returns list of dataframes with useful info 
fit3 <- lapply(names(Dropcolmax), function(x){ 
    tidy(lm(lagcolmean[[x]] ~ Dropcolmax[[x]])) }) 

fit3 

# [[1]] 
#    term estimate std.error statistic p.value 
# 1  (Intercept) -0.4818854 0.1430076 -3.369648 0.1836572 
# 2 Dropcolmax[[x]] 0.5657314 0.1750100 3.232566 0.1909950 
# 
# [[2]] 
#    term estimate std.error statistic p.value 
# 1  (Intercept) -1.4705816 2.7481718 -0.5351127 0.6872022 
# 2 Dropcolmax[[x]] 0.3886215 0.7464919 0.5205971 0.6944294 
# 
# [[3]] 
#    term estimate std.error statistic p.value 
# 1  (Intercept) 1.4288545 0.13717192 10.416523 0.06092962 
# 2 Dropcolmax[[x]] -0.2370706 0.02398423 -9.884433 0.06418790 


# collapse the list to a big dataframe 
dt = do.call(rbind, fit3) 

dt 

#    term estimate std.error statistic p.value 
# 1  (Intercept) -0.4818854 0.14300765 -3.3696477 0.18365719 
# 2 Dropcolmax[[x]] 0.5657314 0.17501003 3.2325659 0.19099495 
# 3  (Intercept) -1.4705816 2.74817181 -0.5351127 0.68720222 
# 4 Dropcolmax[[x]] 0.3886215 0.74649192 0.5205971 0.69442940 
# 5  (Intercept) 1.4288545 0.13717192 10.4165232 0.06092962 
# 6 Dropcolmax[[x]] -0.2370706 0.02398423 -9.8844330 0.06418790 


# returns list of dataframes with useful info and the name of the variable 
fit4 <- lapply(names(Dropcolmax), function(x){ 
    dd = tidy(lm(lagcolmean[[x]] ~ Dropcolmax[[x]])) 
    data.frame(name = x, dd)}) 

fit4 

# [[1]] 
# name   term estimate std.error statistic p.value 
# 1 MSFT  (Intercept) -0.4818854 0.1430076 -3.369648 0.1836572 
# 2 MSFT Dropcolmax[[x]] 0.5657314 0.1750100 3.232566 0.1909950 
# 
# [[2]] 
# name   term estimate std.error statistic p.value 
# 1 AAPL  (Intercept) -1.4705816 2.7481718 -0.5351127 0.6872022 
# 2 AAPL Dropcolmax[[x]] 0.3886215 0.7464919 0.5205971 0.6944294 
# 
# [[3]] 
# name   term estimate std.error statistic p.value 
# 1 GOOGL  (Intercept) 1.4288545 0.13717192 10.416523 0.06092962 
# 2 GOOGL Dropcolmax[[x]] -0.2370706 0.02398423 -9.884433 0.06418790 


# collapse the list to a big dataframe 
dt = do.call(rbind, fit4) 

dt 

# name   term estimate std.error statistic p.value 
# 1 MSFT  (Intercept) -0.4818854 0.14300765 -3.3696477 0.18365719 
# 2 MSFT Dropcolmax[[x]] 0.5657314 0.17501003 3.2325659 0.19099495 
# 3 AAPL  (Intercept) -1.4705816 2.74817181 -0.5351127 0.68720222 
# 4 AAPL Dropcolmax[[x]] 0.3886215 0.74649192 0.5205971 0.69442940 
# 5 GOOGL  (Intercept) 1.4288545 0.13717192 10.4165232 0.06092962 
# 6 GOOGL Dropcolmax[[x]] -0.2370706 0.02398423 -9.8844330 0.06418790 


# get average of intercepts 
mean(dt$estimate[dt$term == "(Intercept)"]) 
[1] -0.1745375