Data-driven testing the fit of linear models
- Spokoiny, Vladimir
2010 Mathematics Subject Classification
- 62H25 62G10
- data-driven test, Haar basis, linear hypothesis, nonparametric alternative, regression model
The paper is concerned with the problem of testing a linear hypothesis about regression function. We propose a new testing procedure based on the Haar transform which is adaptive to unknown smoothness properties of the underlying function. The results show rate optimality of this procedure under mild conditions on the model.
- Math. Methods Statist., 10, no. 4 (2001), pp. 465 - 497