.- help for ^warpdenm^ [STB-38: snp13; STB-53: sg128] .- WARPing density estimation -------------------------- ^warpdenm^ varname [^if^ exp] [^in^ range], ^b^width^(^#^)^ ^m^val^(^#^)^ ^k^ercode^(^#^)^ [^st^tep ^numo^des ^mo^des ^np^oints ^g^en^(^denvar midvar^)^ ^nog^raph graph_options] Description ----------- This program estimates univariate density estimator by means of the ASH-WARPing procedure (Scott, 1985, 1992; Haerdle, 1991), draws the result and provide modality information. Options ------- ^bwidth(^#^)^ is the smoother parameter h (binwidth for histograms, frequency polygons and averaged shifted histograms or FP-ASH; bandwidth for kernel density estimators). ^mval(^#^)^ is the number of averaged shifted histograms used to calculate the required density estimations. ^kercode(^#^)^ specifies the weight function (kernel) to calculate the univariate densities according to the following numerical codes: 1 = Uniform 2 = Triangle 3 = Epanechnikov 4 = Quartic (Biweight) 5 = Triweight 6 = Gaussian ^step^ is included to draw the step (histogram like) version. The default is the linear interpolated (polygon) version. ^numodes^ display the number of modes in the density estimation. ^modes^ lists the estimated values for each modes. The ^numo^des option must be included first. ^npoints^ gives the number of points used for estimation. ^gen(^denvar midvar^)^ permits to create two variables containing respectively the estimated density and the corresponding midpoints used for calculation. ^nograph^ suppresses the graph drawing. graph_options are any of the options allowed with ^graph, twoway^. Remarks ------- ^bwidth^, ^mval^, and ^kercode^ are not optional. If the user does not provide them, the program halts and displays an error message on screen. This program is an all Stata command. In contrast with ^warpden^ it does not require any external executable file. The user is warned that this implementation is not as fast as warpden, specially with a high value for M (a value of 10 is suggested but not obligatory). The "smoothness" of the resulting estimate can be regulated by changing the bandwidth: wide intervals produce smooth results; narrow intervals give noiser results. Except for the Gaussian all the weight functions are supported on [-1,1]. As ^mval^ increases, the approximation is closer to the true kernel estimation, but the quantity of calculation increases too. A good compromise is to use an ^mval^ around 10 (Haerdle, 1991). This procedure can be regarded as a descriptive smoother of histograms besides a nonparametric density estimator. Examples -------- . ^warpdenm snow, bwidth(10) mval(1) kercode(2) step^ Will display a histogram for ^snow^ using a bindwidth of 10. . ^warpdenm snow, b(10) m(1) k(2)^ Will display a frequency polygon for ^snow^. . ^warpdenm snow, b(10) m(5) k(2) step^ Will display the estimate from averaging five histograms with the triangle weight function. . ^warpdenm bodlen, b(10) m(15) k(4) gen(denq10 mid10) nog^ Will calculate the WARPing approximation for the Quartic kernel, and will generate two variables with the resulting density estimation and the corresponding midpoints, without any graphical display. . ^warpdenm snow, b(3) m(10) k(6) numo mo np nog^ Will display the number of modes and their corresponding estimates; besides reports the number of points used to estimate density. Authors ------- Isaias H. Salgado-Ugarte(*), Makoto Shimizu, and Toru Taniuchi (*)Universidad Nacional Autonoma de Mexico Facultad de Estudios Superiores Zaragoza Biologia, Guelatao 66 Col. Ejercito de Oriente Iztapalapa 09230, Mexico D.F., Mexico (Fax 52-5-773-0151) e-mail: fes01@@tzetzal.dcaa.unam.mx; isalgado@@servidor.unam.mx Also see -------- STB: snp6 (STB-16); snp6.1 (STB-26); snp6.2 (STB-27) On-line: ^help^ for @warpdens@, @kerneld@, @bandw@, @l2cvwarp@, @bcvwarp@, @kernreg@, @numode@