Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Non-parametric regression for hypothesis testing using "npregress" in STATA 15

    Firstly, let me accept that I am new to STATA and willing to learn non parametric regression slowly.

    I am trying to employ non-parametric regression for my current research work titled "the impact of infrastructure on manufacturing output in Indian perspective". I estimated the model (as given below) using "npregress" available in STATA 15.


    Model: Output = f(K, Input, Emp, Electricity)

    where, Output, K, Input, Emp and Electricity represent manufacturing output, capital, total input, manpower involved and electricity generation capacity per capita.

    Stata command: npregress kernel Output K Input Emp Electricity, vce(bootstrap, reps(100) seed(111))


    Click image for larger version

Name:	New Bitmap Image.bmp
Views:	1
Size:	1.53 MB
ID:	1454101







    Firstly, How to interpret the current results?

    Secondly, I wanted to know whether "npgraph" works in the multivariate framework or not? I got an error like this: npgraph is only valid with one covariate.

    Thirdly, literature suggest logarithmic transformation of output, capital, manpower, input and infrastructure variables. Whether, non-parametric results will take care of logarithmic transformation of the variables on their own?


    Last edited by Amitshankar Vid; 19 Jul 2018, 01:50.

  • #2
    Stata personnel have written 2 blogs about -npregress-; if you haven't read them, you should start there:

    https://blog.stata.com/2017/06/27/no...ssion-but-not/
    https://blog.stata.com/2018/06/18/ex...ession-models/

    Comment

    Working...
    X