R shows few columns as insignificant for the same feature dummies

ravikanth.karedla
Posts: 5
Joined: Tue May 09, 2017 3:54 am

R shows few columns as insignificant for the same feature dummies

Postby ravikanth.karedla » Mon Jun 19, 2017 5:47 pm

Hi Team,

I'm trying to build multiple linear regression. One of the feature has 22 categories and I've created 21 Dummies. and when build the model R shows few column are significant(***) and few as () for the same 21 dummy variable. Do we need to remove all 21 or what decision do we need to take.

From below make_Dummy 1 till 21 refers same column dummies, however make_Dummy2,5,21 are significant are others are not, so do we need to consider only these 3 features or as it doesn't make sense we need to remove all the 21 dummy features pls help
Residuals:
Min 1Q Median 3Q Max
-1.47607 -0.26091 -0.00282 0.29346 1.64594

Coefficients: (2 not defined because of singularities)
Estimate Std. Error t value Pr(>|t|)
(Intercept) 9.469e+00 5.493e+00 1.724 0.086877 .
normalized.losses 1.540e-03 1.149e-03 1.340 0.182290
make_Dummy1 -5.062e-01 7.059e-01 -0.717 0.474442
make_Dummy2 2.226e+00 5.848e-01 3.807 0.000207 ***
make_Dummy3 3.523e-01 5.525e-01 0.638 0.524668
make_Dummy4 2.182e-01 6.618e-01 0.330 0.742142
make_Dummy5 3.096e+00 6.507e-01 4.759 4.64e-06 ***
make_Dummy6 1.110e+00 7.295e-01 1.522 0.130115
make_Dummy7 1.182e+00 9.303e-01 1.270 0.205977
make_Dummy8 4.747e-01 6.581e-01 0.721 0.471833
make_Dummy9 6.346e-01 1.133e+00 0.560 0.576313
make_Dummy10 7.671e-01 5.883e-01 1.304 0.194294
make_Dummy11 1.154e+00 6.689e-01 1.724 0.086738 .
make_Dummy12 1.283e+00 8.887e-01 1.444 0.150903
make_Dummy13 -3.284e-01 7.972e-01 -0.412 0.681017
make_Dummy14 6.339e-01 5.868e-01 1.080 0.281829
make_Dummy15 -1.126e+00 8.789e-01 -1.281 0.202137
make_Dummy16 1.614e-01 6.688e-01 0.241 0.809635
make_Dummy17 7.603e-01 7.385e-01 1.029 0.304977
make_Dummy18 4.371e-01 6.592e-01 0.663 0.508298
make_Dummy19 1.532e-01 7.626e-01 0.201 0.841088
make_Dummy20 6.906e-01 7.336e-01 0.941 0.348053
make_Dummy21 2.405e+00 7.023e-01 3.425 0.000799 ***
fuel_type_Dummy 3.397e+00 1.953e+00 1.740 0.083986 .
aspiration_Dummy 6.066e-01 2.716e-01 2.234 0.027036 *
num_of_doors_Dummy -2.848e-01 7.561e-02 -3.767 0.000239 ***
body_style_Dummy1 -5.170e-01 4.302e-01 -1.202 0.231451
body_style_Dummy1.1 -7.329e-01 3.933e-01 -1.864 0.064394 .
body_style_Dummy1.2 -5.307e-01 3.701e-01 -1.434 0.153802
body_style_Dummy1.3 -3.444e-01 3.962e-01 -0.869 0.386128
drive_wheels_Dummy1 7.267e-01 3.689e-01 1.970 0.050780 .
drive_wheels_Dummy2 4.520e-02 2.899e-01 0.156 0.876317
engine_location_Dummy -4.158e-01 9.052e-01 -0.459 0.646683
wheel.base -7.535e-02 2.995e-02 -2.516 0.012962 *
length -2.093e-02 1.596e-02 -1.312 0.191746
width 3.667e-02 7.552e-02 0.486 0.627966
height -7.034e-02 4.547e-02 -1.547 0.124050
curb.weight 2.596e-05 5.167e-04 0.050 0.959997
engine_type_Dummy1 9.077e-01 8.898e-01 1.020 0.309338
engine_type_Dummy2 -3.924e-01 4.008e-01 -0.979 0.329156
engine_type_Dummy3 NA NA NA NA
engine_type_Dummy4 -6.655e-01 2.973e-01 -2.238 0.026702 *
engine_type_Dummy5 -1.338e-01 8.383e-01 -0.160 0.873419
engine_type_Dummy6 6.709e-01 1.205e+00 0.557 0.578504
num_of_cylinders_Dummy -4.432e-01 2.016e-01 -2.198 0.029514 *
engine.size 2.993e-02 8.048e-03 3.720 0.000284 ***
fuel_system_Dummy1 1.222e+00 8.886e-01 1.375 0.171152
fuel_system_Dummy2 9.744e-01 6.081e-01 1.602 0.111227
fuel_system_Dummy3 4.975e-01 5.234e-01 0.951 0.343381
fuel_system_Dummy4 2.136e+00 8.754e-01 2.440 0.015893 *
fuel_system_Dummy5 NA NA NA NA
fuel_system_Dummy6 2.525e-01 8.522e-01 0.296 0.767446
fuel_system_Dummy7 6.392e-01 4.979e-01 1.284 0.201231
bore -5.229e-01 4.928e-01 -1.061 0.290409
stroke -4.279e-02 2.989e-01 -0.143 0.886352
compression.ratio 2.358e-01 1.380e-01 1.709 0.089539 .
horsepower -1.766e-02 6.814e-03 -2.591 0.010527 *
peak.rpm 5.150e-05 2.168e-04 0.238 0.812587
city.mpg -5.975e-03 4.486e-02 -0.133 0.894222
highway.mpg 3.033e-03 3.862e-02 0.079 0.937508
price 1.342e-05 2.508e-05 0.535 0.593378
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

Residual standard error: 0.5871 on 146 degrees of freedom
Multiple R-squared: 0.8409, Adjusted R-squared: 0.7777
F-statistic: 13.31 on 58 and 146 DF, p-value: < 2.2e-16

Tags:

edupristine
Finance Junkie
Posts: 778
Joined: Wed Apr 09, 2014 6:28 am

Re: R shows few columns as insignificant for the same feature dummies

Postby edupristine » Fri Jul 21, 2017 9:53 am

Hi Ravi,

A general approach in modeling is to create a model that is simplistic and with minimum number of variables. In the case you mentioned, we do not ignore the variable completely as some levels of it are significant predictors of our dependent variable. We will considers the dummies that are significant according to P-value and ignore the remaining levels. You may come across scenarios where a variable is not significant as per Regression output, but it is an important contributor to predict the target variable as per business understanding. In such case we should not ignore the variable. Revisintg the data and identifying appropriate model are the next steps.


Return to “Business Analytics”



cron

Disclaimer

Global Association of Risk Professionals, Inc. (GARP®) does not endorse, promote, review or warrant the accuracy of the products or services offered by EduPristine for FRM® related information, nor does it endorse any pass rates claimed by the provider. Further, GARP® is not responsible for any fees or costs paid by the user to EduPristine nor is GARP® responsible for any fees or costs of any person or entity providing any services to EduPristine Study Program. FRM®, GARP® and Global Association of Risk Professionals®, are trademarks owned by the Global Association of Risk Professionals, Inc

CFA Institute does not endorse, promote, or warrant the accuracy or quality of the products or services offered by EduPristine. CFA Institute, CFA®, Claritas® and Chartered Financial Analyst® are trademarks owned by CFA Institute.

Utmost care has been taken to ensure that there is no copyright violation or infringement in any of our content. Still, in case you feel that there is any copyright violation of any kind please send a mail to abuse@edupristine.com and we will rectify it.