def test_interp_fill(self): result_lin = masked_array(self.data).astype(float) result_lin[0] = masked test = interp_masked1d(self.test_array.astype(float), kind='linear') assert_almost_equal(test, result_lin)
else: valo2[i1] = float(elem) valo = numpy.array(valo2) dato = [x['date'] for x in resu['observations']['observation']] if freqf == 'D': tso = ts.time_series(valo,dates=dato,freq='D') tso2 = ts.convert(tso,'M',func=numpy.ma.mean) elif freqf == 'W': tso = ts.time_series(valo,dates=dato,freq='D') tso2 = ts.convert(tso,'M',func=numpy.ma.mean) elif freqf == 'M': tso = ts.time_series(valo,dates=dato,freq='M') tso2 = copy.deepcopy(tso) elif freqf == 'Q': tso = ts.time_series(valo,dates=dato,freq='Q') tso2 = ts.convert(tso,'M') tso2 = interp_masked1d(tso2) valor = tso2['1985-01-01':'2011-09-01'] dator = valor.dates valor = valor.data valor = numpy.reshape(valor,(len(valor),1)) matd = numpy.hstack((matd,valor)) tmpstr = tmpstr + r"\bottomrule"+"\n" tmpstr = tmpstr + r"\multicolumn{7}{l}{{\bf Notes:} Transformation codes are: 1=levels, 2=first seasonal difference, 3=second seasonal difference, 4=log level, 5=log first seasonal difference, 6=log second seasonal difference,}\\"+"\n" tmpstr = tmpstr + r"\multicolumn{7}{l}{7=log hp-filtered monthly data. Note that investment data were intrapolated in order to obtain monthly from quarterly data. The choice of fast/slow variables in this table reflects the benchmark ordering.}"+"\n" tmpstr = tmpstr + r"\end{tabular}"+"\n" tmpstr = tmpstr + r"\end{sidewaystable}"+"\n" ''' if 'table1.tex' in os.listdir('../tex_papers/bloom_favar/'): os.remove('../tex_papers/bloom_favar/table1.tex')
else: valo2[i1] = float(elem) valo = numpy.array(valo2) dato = [x['date'] for x in resu['observations']['observation']] if freqf == 'D': tso = ts.time_series(valo, dates=dato, freq='D') tso2 = ts.convert(tso, 'M', func=numpy.ma.mean) elif freqf == 'W': tso = ts.time_series(valo, dates=dato, freq='D') tso2 = ts.convert(tso, 'M', func=numpy.ma.mean) elif freqf == 'M': tso = ts.time_series(valo, dates=dato, freq='M') tso2 = copy.deepcopy(tso) elif freqf == 'Q': tso = ts.time_series(valo, dates=dato, freq='Q') tso2 = ts.convert(tso, 'M') tso2 = interp_masked1d(tso2) valor = tso2['1985-01-01':'2011-09-01'] dator = valor.dates valor = valor.data valor = numpy.reshape(valor, (len(valor), 1)) matd = numpy.hstack((matd, valor)) tmpstr = tmpstr + r"\bottomrule" + "\n" tmpstr = tmpstr + r"\multicolumn{7}{l}{{\bf Notes:} Transformation codes are: 1=levels, 2=first seasonal difference, 3=second seasonal difference, 4=log level, 5=log first seasonal difference, 6=log second seasonal difference,}\\" + "\n" tmpstr = tmpstr + r"\multicolumn{7}{l}{7=log hp-filtered monthly data. Note that investment data were intrapolated in order to obtain monthly from quarterly data. The choice of fast/slow variables in this table reflects the benchmark ordering.}" + "\n" tmpstr = tmpstr + r"\end{tabular}" + "\n" tmpstr = tmpstr + r"\end{sidewaystable}" + "\n" ''' if 'table1.tex' in os.listdir('../tex_papers/bloom_favar/'): os.remove('../tex_papers/bloom_favar/table1.tex') f1 = open('../tex_papers/bloom_favar/table1.tex','w')