linear regression for time series in r- efficiently

Customer  date       sales1 sales2
c1        2019-01-01   67   35
c1        2019-01-07   70   32
c1        2019-01-14   72   40
c2        2019-01-01   100  12
c2        2019-01-07   134  20
c2        2019-01-14   174  23

Making date column as number for forecasting purpose.

df <- df %>% group_by(customer) %>% mutate(dt_seq = row_number())
n <- data.frame()
for(i in unique(df$customer)){
  one <- df[df$customer ==i,]
  one$customer <- NULL
  model <- lm(sales1 ~ dt_seq,data=one)
  fin <- data.frame(matrix(0,52,1))
  colnames(fin) <- 'dt_seq'
  fin$dt_seq <- seq(max(one$dt_seq)+1,max(one$dt_seq)+52,1)
  pre <- as.data.frame(predict(model,fin))
  temp <- cbind(cbind(fin,pre),i)
  temp$dt <- seq(max(one$dt)+7,max(one$dt)+7*52,7)
  colnames(temp) <- c("dt_seq","sales1","customer","dt")
  n <- rbind(n,temp)
}

This is taking long time as I have many customers data and is there any other way to run parallel using spark.lapply() function. Edit : I would like to predict for future values for next 52 periods.For each customer the for loop is taking too long time as I have many customers data. Is there other way to calculate linear regression for time series for grouped data and forecast values.