Linear regression looping in r with alteration in the variable y

Question

I’m doing a regression model Y ~ X1 + X2.

Y are financial market shares (change in daily liquidity).

X1 is a portfolio with the most traded shares (1st quartile), tb with the variation in daily liquidity, but this variation is in the portfolio.

X2 is the market portifolio (all shares).

When Y is inside X1, need to withdraw Y of this portifolio and run the regression in the above model, but in the portfolio X1, Y will not be.

How can I do this in looping, since I have on average 50 shares that are part of the portfolio X1, and for every action, I’ll actually be doing a regression with a X1 different ?

X1 is the last column, sum of actions:

X2 is the last column, sum of actions:

To make the regression:

Petr4 ~ X1 + X2 

I need to get the Petr4 from within the X1. And so, successively, with all who are in X1.

My base has 350 shares and, on average, 50 are part of X1, in each of the 20 quarters analysed.

X1 <- structure(list(ABEV3 = c(-1.68961360937539, 0.336976612313896, 0.493514089699225, 0.254432823660535, -0.256405265540702, 0.264718777827759, 1.55834609347509, -1.20767246429255, -0.141477780693152, -0.19214164080899, -0.000600037380238257, -0.147233871179818, 1.75915399473419, -2.94440084531756, 2.34297132404171, -2.05898734280988, 2.59824344556308, -2.55640564217954, 0.930253209395329), 
B3SA3 = c(-0.249394879240838, -0.435826887737834, 0.674263712359221, -1.47663475639016, 1.65555693916285, -0.577034735765758, 2.26875789306363, -2.2875050945378, 0.537969408805017, -0.205268668573947, 0.414868128804647, -1.48457772878109, 1.15631736949281, 0.0305113207811356, -0.102657510980629, 1.37400357417165, -2.07665582016, 2.98843329137188, -2.69685509115329), 
BBDC4 = c(-0.212217934317245, 0.118614534704256, 0.506732536312761, 0, 0, 1.2116182781466, 1.56667948677674, -3.59090513583332, 2.58117667166673, -2.25835436838038, -0.0164787028530569, -0.475048661790092, 0.120260328263065, 0.69979767549019, -1.00732194106152, 1.39438438911175, -1.58704315346786, 2.26692551507724, 1.01733376126224), 
PETR4 = c(-1.30694127859772, 0.0556754000556631, 1.6427228959543, -2.07784831693627, 0.48951225216644, 1.77717497076481, -0.651276784435446, 0.216236137690635, -0.921123800566615, -0.675165953169317, -0.00202368123748712, -0.0428920777198445, 0.121292232920353, 1.23851717771871, -0.683468222178768, -0.908243883108815, -0.107679168050808, 1.83291659681442, -1.19385481066669), 
VALE3 = c(2.41401127732222, -2.76421566679027, 0.461422210143545, -0.875330431707249, 0.885104981630114, -0.843726971357561, 1.5682910541085, -0.571164168008786, -0.10252567045643, -0.989403988444739, -0.692572450320219, 0.982267519085984, 1.26078160982549, -0.466576892153469, 0.432368421707883, -2.05782217710974, 1.50754449280209, -0.916145594562414, -0.0527556012964956), 
X1 = c(-1.04415642420898, -2.68877600745429, 3.77865544446905, -4.17538068137314, 2.7737689074187, 1.83275031961585, 6.31079774298852, -7.44101072498182, 1.95401882875555, -4.32033461937738, -0.296806742986354, -1.16748482038486, 4.41780553523591, -1.44215156348099, 0.981892071528681, -2.25666543974503, 0.334409796686495, 3.61572416652159, -1.9958785324589)), 
row.names = c(NA, -19L), class = c("tbl_df", "tbl", "data.frame"))


X2 <- structure(list(AALR3 = c(-0.206362587327738, 1.61895166368979, -0.861370994204189, 0.174830746289519, 0.407773465953981, 0.970959829160052, -1.17300376052144, -0.575620575876884, 0.197278912037272, 0.897302548964069, -1.32420067203198, 3.95899769222222, -1.21163567633219, -1.83565877386244, 1.54036374813566, -1.82397761281091, 1.20959432595504, -0.0759481540437981, -0.638232551464981), ABCB10 = c(-1.60296610096249, 2.09620731537456, -0.41694352576488, -0.838618382137318, -0.216899651453602, -0.235666655708187, 4.2059275953516, -2.48173025488413, 4.15244771238538, -2.20385731514288, 0.647168032961616, -4.53221518198161, 2.62619079447456, 0.0935951716704986, -0.354716323827452, 3.91837590914364, -1.93979523690335, 1.61498476891751, 0.122212912954949), 
ABCB4 = c(-1.60296610096249, 2.09620731537456, -0.41694352576488, -0.838618382137318, -0.216899651453602, -0.235666655708187, 4.2059275953516, -2.48173025488413, 4.15244771238538, -2.20385731514288, 0.647168032961616, -4.53221518198161, 2.62619079447456, 0.0935951716704986, -0.354716323827452, 3.91837590914364, -1.93979523690335, 1.61498476891751, 0.122212912954949), 
ABRE11 = c(-1.27341135842787, -1.86497998476113, -0.484129996891839, 1.47010984330541, -0.0210150840181142, -0.357694757179082, 0, 0, 1.62240834012385, 2.80846409804994, -2.58178098976869, 1.15726791232328, -4.12912917187979, 0.695814197684221, 1.9304416196954, -0.687599921266767, 0.0592175350156154, 0.0742655438637557, 0.0469482393875812 ), 
AEDU3 = c(1.22237903311943, -0.0432075917354171, -0.151354670240687, 0.206910952878326, 1.03072282359242, -0.311151317468535, -2.7112186212168, 0.216238189442155, 0.307723724513601, 0.534775159411636, 0.976633521829294, -1.44257539205824, 0.177300976652729, 0.585653957384092, -0.862543274976379, -0.663875728078051, 0.426836767805498, 1.87185456856346, -2.35992132953569 ), 
ABEV3 = c(-1.68961360937539, 0.336976612313896, 0.493514089699225, 0.254432823660535, -0.256405265540702, 0.264718777827759, 1.55834609347509, -1.20767246429255, -0.141477780693152, -0.19214164080899, -0.000600037380238257, -0.147233871179818, 1.75915399473419, -2.94440084531756, 2.34297132404171, -2.05898734280988, 2.59824344556308, -2.55640564217954, 0.930253209395329 ), 
B3SA3 = c(-0.249394879240838, -0.435826887737834, 0.674263712359221, -1.47663475639016, 1.65555693916285, -0.577034735765758, 2.26875789306363, -2.2875050945378, 0.537969408805017, -0.205268668573947, 0.414868128804647, -1.48457772878109, 1.15631736949281, 0.0305113207811356, -0.102657510980629, 1.37400357417165, -2.07665582016, 2.98843329137188, -2.69685509115329 ), 
BBDC4 = c(-0.212217934317245, 0.118614534704256, 0.506732536312761, 0, 0, 1.2116182781466, 1.56667948677674, -3.59090513583332, 2.58117667166673, -2.25835436838038, -0.0164787028530569, -0.475048661790092, 0.120260328263065, 0.69979767549019, -1.00732194106152, 1.39438438911175, -1.58704315346786, 2.26692551507724, 1.01733376126224), 
PETR4 = c(-1.30694127859772, 0.0556754000556631, 1.6427228959543, -2.07784831693627, 0.48951225216644, 1.77717497076481, -0.651276784435446, 0.216236137690635, -0.921123800566615, -0.675165953169317, -0.00202368123748712, -0.0428920777198445, 0.121292232920353, 1.23851717771871, -0.683468222178768, -0.908243883108815, -0.107679168050808, 1.83291659681442, -1.19385481066669), 
VALE3 = c(2.41401127732222, -2.76421566679027, 0.461422210143545, -0.875330431707249, 0.885104981630114, -0.843726971357561, 1.5682910541085, -0.571164168008786, -0.10252567045643, -0.989403988444739, -0.692572450320219, 0.982267519085984, 1.26078160982549, -0.466576892153469, 0.432368421707883, -2.05782217710974, 1.50754449280209, -0.916145594562414, -0.0527556012964956), 
X2 = c(-4.50748353877014, 1.21440271048808, 1.44791273160258, -4.00076590317452, 3.75745081003979, 1.66353076271191, 10.8384305519535, -12.7638536211848, 12.386325230201, -4.48750744323749, -1.9318188170345, -6.55822497186082, 4.50672325262578, -1.80915183893412, 2.88072151672846, 2.40463311638653, -1.84953204834405, 8.71586566274002, -4.70265834816209)), 
row.names = c(NA, -19L ), class = c("tbl_df", "tbl", "data.frame"))

Answer 1

You can do whatever you want without cycles for, cyclically *apply.

First, I’ll redo the base, since it’s easier to have all the columns in the same data.frame.

base <- X1
base$X2 <- X2$X2

Now, I’m going to run the models for each of the answers.

nc <- ncol(base)
respostas <- names(base[-((nc - 1):nc)])

model_list <- lapply(respostas, function(resp){
  fmla <- as.formula(paste(resp, "X1 + X2", sep = "~"))
  lm(fmla, data = base)
})

names(model_list) <- respostas

This result can then be used to extract the information you want.

For example, the regression coefficients:

t(sapply(model_list, coef))

To extract more things, it is better to use summary.

model_smry <- lapply(model_list, summary)

And from the summaries we take the tables with the coefficients we already have, the values of the statistics t or the p.values, among others.
To extract elements from these tables is a little more complicated but the process is always the same, with a little practice becomes intuitive.

model_coef <- lapply(model_smry, '[[', "coefficients")
model_coef <- lapply(model_coef, as.data.frame)
sapply(model_coef, '[[', "Pr(>|t|)")

For example, the coefficients of determination directly from the summaries, since they are not in the tables of coefficients:

sapply(model_smry, '[[', "r.squared")
#    ABEV3     B3SA3     BBDC4     PETR4     VALE3 
#0.1333845 0.6140561 0.5823004 0.1924847 0.5148558