M/Phl&ddlmZmZddlmZddlmZddlZddl m Z ddlZddl m Z ddl Z ddlZddlmZdZdZd Zd ZGd d ZGd dZGddZdZGddZdZdZdZdZGddZGddeZGddeZ GddeZ!Gd d!eZ"Gd"d#eZ#Gd$d%eZ$Gd&d'eZ%Gd(d)eZ&Gd*d+e"Z'dS),)assert_series_equalassert_frame_equal)StringIO)dedentN) assert_equal)rosctd}tj|}|d|jdddf<|ddk|jdddf<|S)Nares,qual 2.00,= 4.20,= 4.62,= 5.00,ND 5.00,ND 5.50,ND 5.57,= 5.66,= 5.75,ND 5.86,= 6.65,= 6.78,= 6.79,= 7.50,= 7.50,= 7.50,= 8.63,= 8.71,= 8.99,= 9.50,ND 9.50,ND 9.85,= 10.82,= 11.00,ND 11.25,= 11.25,= 12.20,= 14.92,= 16.77,= 17.81,= 19.16,= 19.19,= 19.64,= 20.18,= 22.97,= resconcqualNDcensored)rpandasread_csvloc)raw_csvdfs e/var/www/html/test/jupyter/venv/lib/python3.11/site-packages/statsmodels/imputation/tests/test_ros.pyload_basic_datarsb AG  ! !B5 BF111f9vJ$.BF111j= Ictjgdddddddddddddddddddddd d dddd d dddd d ddd ddddd ddddd ddddd ddddd ddddd ddddd ddddd ddddd ddd dd ddd dd ddddd ddddd ddddd ddddd ddd dd d!d ddd d"d ddd d#d ddd d#d ddd d$d ddd d%d d dd d&d d dd d'd ddd d(d ddd d)d ddd d*d ddd d+d d dd d,d d-d}|S).NT@)rr det_limit_indexrank@@#@&@F@r@{Gz@HzG@p= ף@q= ףp@皙@Q@)\(@@(\B!@Qk!@ {G!@ 33333#@p= ף%@&@ffffff(@ףp= -@Q0@(\1@)\(3@q= ף03@p= ף3@Gz.4@Q6@ r DataFramers rload_intermediate_datarF s'  $31aHH$31aHH$31aHH$4AqII $ 31aHH $ 31aHH $4AqII$CAqII$CAqII$DQJJ$DQJJ$DQJJ$DQJJ$DQJJ$DQJJ$ DQJJ!$"CAqII#$$CAqII%$&CAqII'$(DQJJ)$*DQJJ+$,DQKK-$.DQJJ/$0EaKK1$2EaKK3$4EaKK5$6DQJJ7$8EaKK9$:EaKK;$<EaKK=$>EaKK?$@EaKKA$BEaKKC$DEaLLE$FEaLLG$$ $ BL Irc tjgddddddddddddd dd dd d d ddd ddddddddddddddddddd dddddddddddddddddd ddd dd!dd"dd#ddd$dd%d d&ddd'dd(d d)d dd*dd+dd,ddd-dd.dd/d dd0dd1dd2ddd3dd4dd5ddd6dd7dd8ddd9dd7dd:d;dddd?dd@ddAdBddCddDddEdFddGddHddIdJddKddLddMdddNddOddPd ddQddRddSdddTddRddUd ddVddWddXdddYddZdd[ddd\dd]dd^ddd_dd`ddad;ddbddcdddd>ddeddfddgdBddhddiddjdFddkddlddmdJddnddoddpdqd}|S)rNgCB!Trrg$F?)Zprelimrr rplot_posrgXHg6iqx?rgJ@rg$F?gI@rrg$F?gXr r!gs?gW'>ٿg__?gz0zʿr"r#s?g{Fr$rg6iqx?r%g?r&g)YUZ?gL{br'<8 E?gMQnr(g!Rk f?g#ܿr)gGGt)?g!6|Rf׿r*gj` ?g@>(xҿr+gL_?g? $R˿r,gr8-?g@Dr-g+$?g?V$kmg{?r.g.+NƷ?gϋ~K3?r/gȎE?r0g+(?r1g[ T?r2gj?r3go4%?r4g?r5g?r6gU ?g?糼k?r7gyڽN?g?r8g~?g&.Ef?gyy?gXu ?r9gPuPu?gU?r:gKK?gK4?r;g""""""?g?r<g?gM?r=g<g]ʥ\ʥ?g%͢?r?gW|W|?g1_B?r@g.R.R?gj2 ?rAgr)r)?rBrCrEs rload_advanced_datarLJs  G'T3*=q J JG(T3*=q J JG 'D#*=q J J G(T4* K KG"(UC*=q J J#G&(UD*=q J J'G*(UD*=q J J+G.(UD*=q J J/G2)eT*)eT*r?r@rAr r)rrrrDexpected_baselineilocexpected_with_warningselfs r setup_methodzTest__ros_sort.setup_methods!##!'!13  , ,3 =AC0P0P3  , ,3 =AD0Q0Q3  , ,3 >BC0P0P3  - - 3 >CC0P0P 3  , , 3 >CD0Q0Q 3  - - 3 >CD0Q0Q 3  - -3 >CD0Q0Q3  - -3 >CD0Q0Q3  , ,3 >CC0P0P3  , ,3 >CD0Q0Q3  - -3 >CD0Q0Q3  - -3 >CE0R0R3  . .3 >CE0R0R3  - -3 >CE0R0R3  . .3 >CE0R0R3  . .!3 >CE0R0R!3 " . .#3 ">CE0R0R#3 $ . .%3 " " & J '"!*&*%;%@"%E"""rcftj|jdd}t||jdS)Nr r)r _ros_sortrrrjrnresults r test_baselinezTest__ros_sort.test_baselines0tw ;;64#9:::::rc|j}|d}d|j|df<t j|dd}t ||jdS)Nr Tr)rcopyidxmaxrrrqrrl)rnrmax_rowrss rtest_censored_greater_than_maxz-Test__ros_sort.test_censored_greater_than_maxsc W\\^^V*##%%&*w "#r6:6664#=>>>>>rN)__name__ __module__ __qualname__rortryrrrgrgsDFFF4;;;?????rrgc eZdZdZdZdZdS)Test_cohn_numberscnt|_gd|_tjddddddddddddd dd dd ddd dd dd d dddddddddddddddt jdt jt jt jt jdt jdg|j|_dS)N)rRrWrUrTrSrVr$rNrOrPrrQrXrrYrrZr[r\r r]r^r"r_r`) rr final_colsrrDrarbrcrjrms rrozTest_cohn_numbers.setup_methods!##<<<"(!1Cs Sc K KCs 5HVY [ [Cs 5HVZ \ \S!6HVY [ [Ct 5HVZ \ \S!6IW\W` b b%)59"Y3EI W W3 " "  ?"rchtj|jdd}t||jdS)Nr r observations censorship)r cohn_numbersrrrjrrs rrtzTest_cohn_numbers.test_baselines4!$':VVV64#9:::::rc|j}d|d<tj|dd}|jdksJdS)NFr r r)rr.)rrvrrshape)rn_dfrss r test_no_NDszTest_cohn_numbers.test_no_NDssJgllnnF !#FvNNN|v%%%%%%rN)rzr{r|rortrr}rrrrsA.;;;&&&&&rrc&eZdZdZdZdZdZdS)Test__detection_limit_indexct|_tjt jd|_dS)N)rr/)rerdrrDraempty empty_cohnrms rroz(Test__detection_limit_index.setup_methods/#%%  *5;v+>+>??rcVttjd|jddSNr)rr_detection_limit_indexrrms r test_emptyz&Test__detection_limit_index.test_emptys'S/doFFJJJJJrcttjd|jdttjd|jdttjd|jddS)N @rrYr(@r#)rrrrdrms rtest_populatedz*Test__detection_limit_index.test_populatedseS/TY??CCCS/TY??CCCS/di@@!DDDDDrctjt5tjd|jddddS#1swxYwYdSr)pytestraises IndexErrorrrrdrms rtest_out_of_boundsz.Test__detection_limit_index.test_out_of_boundss ]: & & 5 5  &q$) 4 4 4 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5sAA AN)rzr{r|rorrrr}rrrrsY@@@KKKEEE 55555rrc tjdgdztdtdzttdd}t j|dd}tjgdd }t|t|tdS) Nr AABCCCDEDCBA)dl_idxparamsvaluesrr) rrrrrrrrrr!rrr)name) rrDlistranger_ros_group_rankSeriesrastypeint)rrsexpecteds rtest__ros_group_rankrs  #(z""T&\\1uRyy//  B  Xx 8 8F}AAAOOOH c**HOOC,@,@AAAAArc,eZdZdZdZdZdZdZdS)Test__ros_plot_posc,t|_dS)N)rerdrms rrozTest__ros_plot_pos.setup_methods#%% rchdddd}tj|d|j}t|ddS)NFrrrrrrrKr _ros_plot_posrdrrnrowrss rtest_uncensored_1z$Test__ros_plot_pos.test_uncensored_1s> QBB"3 DI>>V011111rchdddd}tj|d|j}t|ddS)NFrrrrg}?rrs rtest_uncensored_2z$Test__ros_plot_pos.test_uncensored_2s> QCC"3 DI>>V011111rchdddd}tj|d|j}t|ddS)NTr#r!rrg__?rrs rtest_censored_1z"Test__ros_plot_pos.test_censored_1!s>AqAA"3 DI>>V/00000rchdddd}tj|d|j}t|ddS)NTr!rrrrJrrs rtest_censored_2z"Test__ros_plot_pos.test_censored_2&s>AqAA"3 DI>>V011111rN)rzr{r|rorrrrr}rrrrs_&&&222 222 111 22222rrctjgd}tjgd}t j||dS)N)rrrr!)gT]?g`>?gO:`?gyj?)r_norm_plot_posraarraynptassert_array_almost_equal)rsrs rtest__norm_plot_posr,sE   - -F{GGGHHH!&(33333rct}t}tj|d|}t jgd}t j||dS)Nr)#g@,U?眤Uqx?^c@?rg?g_?g?gaˀqx?rgB UZ?geL$E?gKn f?g*?gh ?gޮ_?gqo-?g>?g{X={?gK3?g m(?g(?g1^?g?g}N?g"}?gCy?g&(Nu?gC" K?g`C#""?gp?g!Rr?r@rAr rfinal) rarrLr_imputelogexprrrrrrss r test__imputerCso{     H   B RUY B BB [ F!&(33333rctjgd}t}tj|ddtjtj}|dj}tj ||dSr) rarrr_do_rosrrrrrrs r test__do_rosrUso{     H   B RUY B BB [ F!&(33333rc eZdZdZdZdZdS) CheckROSMixinctj|j|j|j}t jt|t|j|j dSNrE)decimal) r impute_rosrescolcencolrrrsortedexpected_finalrrrs r test_ros_dfzCheckROSMixin.test_ros_dfis^ T[TWEEE % 6NN 4& ' 'L      rctj|j|j|j|jd}t jt|t|j|j dSr) rrrrrrrrrrrrs rtest_ros_arrayszCheckROSMixin.test_ros_arraysqsj 4dgdk6JtTTT % 6NN 4& ' 'L      rcgd}tj|j|j|j}t t j||dt j|j|ddS)NrUrTrSrVr) rrrrrrnpround expected_cohn)rncolsrds r test_cohnzCheckROSMixin.test_cohnysu   dkBB HT$Z # # HT'-q 1 1     rN)rzr{r|rrrr}rrrrhsA            rrc eZdZdZdZejgdZejgdZdZ dZ e j e ee eiZ ejgdZe j ejdd ejgejd d ejgejd dejgejgd d Zd S)Test_ROS_HelselAppendixBz Appendix B dataset from "Estimation of Descriptive Statists for Multiply Censored Water Quality Data", Water Resources Research, Vol 24, No 12, pp 1997 - 2004. December 1988. rrPrPrPrPrPrPr[r[r[rO@rZr.@4@;@@@I@)TTTTTTTTTFFFFFFFFFobscen)gGz?g333333?g(\?gRQ?g)\(?gQ@g@rOg= ףp=@g333333@rrZrrrrrrrOrYr)g%?g1ZGU?rNrNrzr{r|__doc__rrarr rrrrrDrrrcrr}rrrrs, G %+  C %+  C F F  634 5 5B U["""N %F$# S#uy$9::!ek3ei"899!ek3UY"788&5;'>'>'>?? &&MMMrrc eZdZdZdZejgdZejgdZdZ dZ e j e ee eiZ ejgdZe j ejdd d d ejgejd d d dejgejd d ddejgejgddZdS)Test_ROS_HelselArsenica! Oahu arsenic data from Nondetects and Data Analysis by Dennis R. Helsel (John Wiley, 2005) Plotting positions are fudged since relative to source data since modeled data is what matters and (source data plot positions are not uniformly spaced, which seems weird) r)皙 @ffffff@r$r$r$r$r$r$r$r$333333??rPrPrPrP?rffffff?r333333??rr)FFTTTTTTTTFFTTTTFTFFFFFFrr)rrgQ?g= ףp=?gffffff?Q?g(\?g= ףp=?gq= ףp?gffffff?rrg\(\?gRQ?g(\?g= ףp=?rgQ?rrrrrrrYrPr$rNrrg6@@ @)rPg?g=m?g&S?rNrNrr}rrrrs8G %+  C %+  C F F  634 5 5B U["""N %F$# S#sC$CDD!ek3T4"CDD!ek3S#uy"ABB&5;'J'J'JKK &&MMMrrc ^eZdZdZeedZdZdZe j edZ e j gdZe je j dd d e jge j d d d e jge j d d de jge j gddZdS)Test_ROS_RNADAdatara res cen 0.090 True 0.090 True 0.090 True 0.101 False 0.136 False 0.340 False 0.457 False 0.514 False 0.629 False 0.638 False 0.774 False 0.788 False 0.900 True 0.900 True 0.900 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 False 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.000 True 1.100 False 2.000 False 2.000 False 2.404 False 2.860 False 3.000 False 3.000 False 3.705 False 4.000 False 5.000 False 5.960 False 6.000 False 7.214 False 16.000 False 17.716 False 25.000 False 51.000 Falser rz\s+)sep)2g6 j?g}%?gM "?gB`"۹?g rh?g(\?gsh|??g r?gT㥛 ?g"~j?gS㥛?gK7?gJ c?πz3*?gN?g&V?g7F?gV?gq C/7?gɹEGϹ?rPgzc?gfD̙+?gc T?rg-?g",(?gG4L?gL6S?g?ΰ(W~?gR*?gZKfd?gJE?g?r$r$gZd;@gzG@rOrOgp= ף @rrgףp= @rYgB`"@g0@gK1@g9@gI@rZrNg2@rOrg@@1@)gzG? ףp= ?rrrN)rzr{r|rrr datastringrrrrrrarrrDrcrr}rrrrsG&&3"3333JhF F  0 0 0B U[ " " "  N%F$# Rc59$=>>!ek2sC";<<!ek2r3 ":;;&5;'<'<'<== &&MMMrrczeZdZdZejddZejeZ dgezZ dZ dZ e je e e e iZejgdZe jejgejgejgejgd Zd S) Test_NoOp_ZeroNDrr)sizeFrr)RQ?Q?rQ?rGz?ffffff?Q?ffffff?ףp= ?)\(?(\?Q@Q@HzG@Q@ףp= @\(\@zG@"@rN)rzr{r|rrarandomseedN lognormalr rrrrrDrrrrr}rrr r sG La A , a ( (C 'A+C F F  634 5 5B U["""N %F$# B!ek"oo!ek"oo&5;r?? &&MMMrr c veZdZdZejgdZejgdZdZdZ e j eee eiZ ejgdZ e j ejdejgejdejgejdejgejd d gd Zd S) Test_ROS_OneNDrr)TFFFFFFFFFFFFFFFFFr r)gQ?rPrPrPrPrPr[r[r[rOrrZrrrrrrrrPg~!<8?rNrNrzr{r|rrarr rrrrrDrrrcrr}rrr%r%2sG %+  C %+  C F F  634 5 5B U["""N %F$# T59$566!ek3 "233!ek3 "233&5;~66 &&MMMrr%ceZdZdZejgdZejgdZdZdZ e j eee eiZ ejgdZ e j ejddddddddejg ejd d d d d dddejg ejd dddddddejg ejdgdzdgzdZdS)Test_HalfDLs_80pctNDsrr)TTTTTTTTTTTTTTTFFFvaluer )rrrrrrrrrrrg@rYr-r[rrrrNrOrYrrrZrg*@g,@rrPgKqU?r1rNr&r}rrr(r(MsJG %+  C %+  C F F  634 5 5B U["""N %F$# RRRRUY$OPP!ek2r2r3S#uy"QRR!ek2r2r2r2r59"MNN&5;y1}t';<< &&MMMrr(c |eZdZdZejgdZejgdZdZdZ e j eee eiZ ejgdZ e j ejddejgejd d ejgejd dejgejgd d Zd S)Test_HaflDLs_OneUncensoredr)rPrPrr)TTTFr)r )rrrYrrNrPr$rO)?r,rNrNr&r}rrr+r+hsG %+,,, - -C %+000 1 1C F F  634 5 5B U[!:!:!:;;N$F$# RUY$788!ek2r59"566!ek2r59"566&5;'7'7'788 &&MMMrr+cVeZdZejgdZejgdZdS)Test_ROS_MaxCen_GT_MaxUncen)rPrPrPrPrPrPr[r[r[rOrrZrrrrrr<F)TTTTTTTTTFFFFFFFFFTTN)rzr{r|rarr rr}rrr.r.ysT %+  C %+  CCCrr.cdeZdZejddZgdZdgezddgzZdS)Test_ROS_OnlyDL_GT_MaxUncenrr )rrrrrrrrrrrrrrrrrrrrr[r[FTN) rzr{r|rar r!r"r rr}rrr2r2sP La A   C 7Q;4, &CCCrr2)(statsmodels.compat.pandasrriortextwraprrar numpy.testingtestingrrrrstatsmodels.imputationrrrFrLrergrrrrrrrrrrrrr r%r(r+r.r2r}rrr9sMMMMMMMM &&&&&& &&&&&&   '''TJJJZ($?$?$?$?$?$?$?$?N & & & & & & & &F55555555$ B B B222222222444 555"444$444&        <}@$$$$$]$$$NKKKKKKKK\}.]6M6"     ":   '''''"2'''''r