Liner regression between 2 variables but only in selected lines.

[ThiagoNM]

Hi everyone,
I am really beginner in R and it is my first day in this community.

I need help to understand how I can define the command to run a linear regression only in specific lines.

For instance, I have a database with several lines and columns: “country” (1, 2, 3, and 4), “year” (1, 2, 3, 4, 5, 6), “period” (1, 2, 3), “advertising ”, and “sales”.

How can I run linear regression between “advertising” and “sales” but using data only there are country 2, year 3, and period 2?

How can I fit that situation in the command for the model below ?
mod <- lm(sales ~ advertising, data)
summary (mod)

The same situation to run a descriptive analysis (mean, min, max, ...) of sales when, for example, it’s county 1 and period 3.

Basically, I would like to create conditions to run analysis only in specific lines based on different columns.
I don’t know if I explained clearly.

Thank you,

[startz]

library(tidyverse)
newData <- data %>%  filter( country == 2 & year == 3 & period == 2)

then use newData for your analysis.

[ThiagoNM]

I really appreciate your reply

If I need to analyze the period 2 for each country (1, 2, 3, and 4), do I need to analyze separated or I can use a code to give me the result for each country ?

[andresrcs]

Take a look at this related thread, this shows how to perform linear regressions over nested subgroups of your data frame

Linear regression by rows

You can use this example as starting point library(dplyr) library(tidyr) library(purrr) library(broom) data <- data.frame(stringsAsFactors=FALSE, ITEM = c("item A", "item B"), JAN = c(10, 30), FEB = c(20, 35), MAR = c(15, 37), APR = c(17, 38) ) fit_model <- function(df) lm(value ~ n, data = df) get_slope <- function(mod) tidy(mod)$estimate[2] data %>% gather('month', 'value', -ITEM) %>% group_by(ITEM) %>% mutate(n = row_number()) %>% arrange(ITEM, n) %>% nest() %>% mutate(model = map(data, fit_model)) %>% mutate(slope = map_dbl(model, get_sl…

If you need more specific help, please provide a proper REPRoducible EXample (reprex) illustrating your issue.

[startz]

@andresrcs gives a link for doing it the "right way" in R. In your case you could simplify by a loop like

for (iCountry in 1:4){
     newData <- data %>%  filter( country == iCountry  & period == 2)
# run the regression or whatever here.
}

[ThiagoNM]

Fantastic! I really appreciate it
I will try to run it

Does it work too if I want to create a graphs for each country using ggplot?

I believe it is very basic, but as I a very beginner, I am struggling with that.
Thank you

[startz]

Sure. The code goes through for each country and pulls out one country at a time. You can run a regression, make a plot, or whatever you like using newData.