I'm currently running a quantile regression model in R using Roger Koenker's quantreg package. I have an ordered categorical variable with five levels and three unordered categorical variables (day of the week, season, and whether or not it's nighttime) with seven, four, and two categories, respectively. All are originally stored in the data frame as strings except the ordered categorical variable, which is five levels from 1 to 5. Here's how I code the previously-mentioned variables as factors. Note that my variable is called df_prime because it's the original data frame with some pruned rows:
df_prime$acuity_id <- ordered(df_prime$acuity_id, c(5, 4, 3, 2, 1))
df_prime$day_of_week <- as.factor(df_prime$day_of_week)
df_prime$season <- as.factor(df_prime$season)
df_prime$is_night <- as.factor(df_prime$is_night)
When I run the regression, it completely cuts Friday from the summary and one of the levels of my acuity_id, which is my ordered variable. Is that possibly related to how R creates dummy variables? I'm noticing that seasonWinter has a p-value of 1 for whatever reason.
My output is as follows, although with different spacing because my workplace has R on a VM and I can't copy and paste:
Value Std. Error t value Pr(>|t|)
(Intercept) 173.20000 5.84510 29.63167 0.00000
day_of_weekMonday 19.33333 3.60107 5.36878 0.00000
day_of_weekSaturday -49.66667 3.41145 -14.55883 0.00000
day_of_weekSunday -42.00000 3.35297 -12.52620 0.00000
day_of_weekThursday 5.33333 3.51831 1.51588 0.12957
day_of_weekTuesday 3.00000 3.51917 0.85247 0.39396
day_of_weekWednesday 1.66667 3.75717 0.44360 0.65734
is_night1 -53.00000 2.04268 -25.94626 0.00000
seasonAutumn 8.00000 4.37771 1.82744 0.06765
seasonSpring -0.66667 4.46209 -0.14941 0.88123
seasonSummer 12.66667 4.40002 2.87877 0.00400
seasonWinter 0.00000 4.36418 0.00000 1.00000
acuity_id.L 37.94733 11.20886 3.38548 0.00071
acuity_id.Q -108.68624 9.48982 -11.45293 0.00000
acuity_id.C -36.36619 5.93905 -6.12324 0.00000
acuity_id^4 25.61773 2.77411 9.23459 0.00000