ݽTf&dZddlZddlmcmZddlZddlm Z ddl m Z m Z m Z mZmZddlmZmZdZdZdZd Zd Zd Zd ZdS) z5 Created on Fri May 10 10:15:23 2024 @author: atdou N)Regression_Outlier_Removal)Linear_RegressionDecision_Tree_RegressionGradient_Boost_RegressionQuadratic_RegressionPolynomial_Regression)create_regress_dataCM_ROR_predictionsc t}d}d\}}|dd||d||ddd |d d |j}|j}t gd d }d|i}t|d ddddd}|||d|j i} t|j | d} tj |dgd||z gg} | | k} | j} | }|stjd| fd| | fdt#jvstj| rtj| nddt#jvstj| rtj| nddz}d|tj| tj|dz}t+tj|dx} x} }dS)z Description ----------- testing that fit method correctly classifies all points using the final iteration (same as average in this case) of a polynomial regressor. First have to create data using create_data class. cld|ddzzd|dzzd|ddzzzd|dzz S N rxs dC:\Users\atdou\OneDrive\Desktop\Files\Coding\Python\Programs\XtraMLTools\tests\test_Preprocessing.pyz"test_ROR_fit_ave..="adQY,1Q4'!AaD!G)3a!f<irrx_minx_maxNn_outdimfuncdevmin_ymax_yIQRmetricfactorx_0x_1x_2 poly_columnsdegreePRGz?r?r.r/n_maxpercentwindow threshold train_aver==z%(py0)s == %(py2)sCM CM_actualpy0py2>assert %(py8)s {%(py8)s = %(py6)s {%(py6)s = %(py4)s.all }() }py4py6py8N)r set_Xset_yclassify_outliersXyrrfittrain_data_aver datanumpyarrayall @pytest_ar_call_reprcompare @py_builtinslocals_should_repr_global_name _safereprAssertionError_format_explanation)data_expfr"r#rPrQr7 model_dictROR data_dictrCrD @py_assert1 @py_assert5 @py_assert7 @py_format3 @py_format9s rtest_ROR_fit_averjsD#$$H<>> NNq"N555 eA666 A A ,A,A,A! L L LBJ $Zar[_hiux y y yCGGAaLLLc01I HM9 5 5k BB eQZAeG 566I""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""rc $t}d\}}d}|dd||d||ddd |d d |j}|j}t gd d }d|i}t|d ddddd}|||| ||d|j i} t|j | d} tj|dgd||z gg} | | k} | j} | }|st!jd| fd| | fdt%jvst!j| rt!j| nddt%jvst!j| rt!j| nddz}d|t!j| t!j|dz}t-t!j|dx} x} }dS)z Description ----------- testing that the transform method correctly classifies all points it just fit, using a polynomial regressor. First have to create data using create_data class. rcld|ddzzd|dzzd|ddzzzd|dzz Sr rrs rrz(test_ROR_transform_ave..1rrrrrrr%r&r'r,r-r0r4r7r8rr9r:test_averr@rBrCrDrErHrIN)r rMrNrOrPrQrrrR transform test_data_aver rTrUrVrWrXrYrZr[r\r]r^r_)r`r"r#rarPrQr7rbrcrdrCrDrerfrgrhris rtest_ROR_transform_averp(sX#$$HHAu<>> NNq"N555 eA666 A A ,A,A,A! L L LBJ $Zar[_hiux y y yCGGAaLLLMM!AS./I HM9 5 5j AB eQZAeG 566I""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""""rc t}d}d\}}|dd||d||ddd |d d |j}|j}t }td g }tgdd}td} td} td} td} tdd} |||| | | | | d}t|d ddddd}| ||| |||jD]4}|j|dd|j|ddk|z }|j}||k}|st'jd|fd ||fd!t+jvst'j|rt'j|nd!d"t+jvst'j|rt'j|nd"t'j|d#z}d$d%|iz}t3t'j|d&x}}6d&S)'a Description ----------- testing that when do ROR.fit(X,y) and get final prediction of outliers, this agrees with the ROR.transform(X,y), to within the percent_necessary threshold - because can't guarantee they'll be the same after every fit. Checking this is true for every model in self.models. If it isn't, then doesn't mean anything is necessarily wrong with program, but wouldn't be good for results. First have to create data using create_data class. cld|ddzzd|dzzd|ddzzzd|dzz Sr rrs rrz2test_ROR_fit_and_transform_dicts..Lrrrrrrrr%r&r'r,r-r1 quad_columnsr0r4r)min_samples_leafrr2{Gz?ru ccp_alpha)LRQRr7DTR_0DTR_1DTR_2DTR_3GBRr8rr9r:Outlying_Predictionr)>)z8%(py0)s > %(py4)s {%(py4)s = %(py2)s.percent_necessary } percent_matchrc)rFrGrJzassert %(py6)srKN)r rMrNrOrPrQrrrrrrrRrnmodels train_data test_datasumpercent_necessaryrXrYrZr[r\r]r^r_)r`rar"r#rPrQrzr{r7r|r}r~rrrbrckeyr @py_assert3re @py_format5 @py_format7s r test_ROR_fit_and_transform_dictsrBs#$$H<>> NNq"N555 eA666 A A   B E7 3 3 3B ,A,A,A! L L LB $a 8 8 8E $a 8 8 8E $b 9 9 9E $b 9 9 9E #Q$ G G GC"BY^inwz{{J $Zar[_hiux y y yCGGAaLLLMM!Az55,R01FG3=Y\K]^_K`avKww||~~@AA 4444444444444444444444444444444444444444444444444444444444444444444444444444444444444444455rc Zt}d}|ddddd||ddd |d d |j}|j}t }td g}tgdd}tdd}tdd}tdd} tdd} tdd} |||||| | | d} t| d ddddd} | ||dddS)z Just testing that cross_val_hist executes properly, outputting histograms for each of the folds and models, can't say if they're correct. cld|ddzzd|dzzd|ddzzzd|dzz Sr rrs rrz)test_ROR_cross_val_hist..irrrrrr(r'r,r-r1rsr0r4-C6?rxrw皙?rrzr{r7r}r~rDTR_4rr%r8rr9r:r)n_foldsn_binsN) r rMrNrOrPrQrrrrrrcross_val_histr`rarPrQrzr{r7r}r~rrrReg_Dictrcs rtest_ROR_cross_val_histrdsb#$$H<.rrrrrrrrrrr'r,r-r1rsr0r4rrxrwrrrr%r8rr9r:N) r rMrNrOrPrQrrrrrrcross_val_scoresrs rtest_ROR_cross_val_scoresr|s[#$$H<.rrrrrrrrrrr'r,r-r1rsr0r4rrxrwrrrr%r8rr9r:rrr7rr7r)rN)r rMrNrOrPrQrrrrrrrRrres_histr`rarPrQrzr{r7r}r~rrrrrcrds rtest_ROR_res_histrs#$$H<.rrrrrrrrrrr'r,r-r1rsr0r4rrxrwrrrr%r8rr9r:rrr7rN)r rMrNrOrPrQrrrrrrrRroutlier_overlaprs rtest_ROR_outlier_overlaprs#$$H<rs&  @@@@@@QQQQQQQQQQQQQQEEEEEEEE###2###4555D33300'''4#####r