NOxEmissions       package:robustbase       R Documentation(latin1)

_N_O_x _A_i_r _P_o_l_l_u_t_i_o_n _D_a_t_a

_D_e_s_c_r_i_p_t_i_o_n:

     A typical medium sized environmental data set with hourly
     measurements of NOx pollution content in the ambient air.

_U_s_a_g_e:

     data(NOxEmissions)

_F_o_r_m_a_t:

     A data frame with 8088 observations on the following 4 variables.

     '_j_u_l_d_a_y' day number, a factor with levels '373' ... '730',
          typically with 24 hourly measurements.

     '_L_N_O_x' log of hourly mean of NOx concentration in ambient air
          [ppb] next to a highly frequented motorway.

     '_L_N_O_x_E_m' log of hourly sum of NOx emission of cars on this
          motorway in arbitrary units.

     '_s_q_r_t_W_S' Square root of wind speed [m/s].

_D_e_t_a_i_l_s:

     The original data set had more observations, but with missing
     values. Here, all cases with missing values were omitted
     ('na.omit(.)'), and then only those were retained that belonged to
     days with at least 20 (fully) observed hourly measurements.

_S_o_u_r_c_e:

     Ren Locher, E-mail to R-SIG-robust mailing list, on 2006-04-20.

_E_x_a_m_p_l_e_s:

     data(NOxEmissions)
     plot(LNOx ~ LNOxEm, data = NOxEmissions, cex = 0.25, col = "gray30")

     ## Not run: ## these take too much time --
      ## p = 340  ==> already Least Squares is not fast
      (lmNOx <- lm(LNOx ~ . ,data = NOxEmissions))
      plot(lmNOx) #->  indication of 1 outlier

      M.NOx <- MASS::rlm(LNOx ~ . , data = NOxEmissions)
      ## M-estimation works
      ## whereas  MM-estimation fails:
      try(MM.NOx <- MASS::rlm(LNOx ~ . , data = NOxEmissions, method = "MM"))
      ## namely because S-estimation fails:
      try(lts.NOx <- ltsReg(LNOx ~ . , data = NOxEmissions))
      try(lmR.NOx <- lmrob (LNOx ~ . , data = NOxEmissions))
     ## End(Not run)

