Thanks to visit codestin.com
Credit goes to www.scribd.com

Scribd
Upload
48 views78 pages

PS - Unit-5

Building LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, FBuilding LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, FBuilding LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, F

Uploaded by

gunayow
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
48 views78 pages

PS - Unit-5

Building LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, FBuilding LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, FBuilding LLMs for Production: Enhancing LLM Abilities and Reliability with Prompting, F

Uploaded by

gunayow
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 78
UNIT - 5 APPLIED STATISTICS Under the test of small zamples , we Shall consider ome — {oltowing important tests are given below 1) t- test for means 2) F- test for yaviantes 3) Chi- squove test for goodness of fit +- TEST FOR SINGLE MEAN are ee H), %2,....%n be the Small samples lie, ne 30) dxawn from the normal population with mean and yovionce 5° then the sample mean |Z. £ y; | and the stondard deviation Het then the test slatistie is © sin | Tf p_be the population mean and n (<30) be a dawn ample mean % avd standard diviation « , then the tect stabisties is b= sin- | 7 The degrees af pusdom of thio slatisties is ven-! Properties OF _t- DISTRIBUTION eee ES OF ee IBUTLCR = 1. The pdb of & distribution is 4ce) - (14 #1) , where to mL eH | Fy de et Sita Al{n-2 Sh 2. The values of +t vanies prom -2 to 0 3.TE has symmetrical about t=o0 and mean xno. 4. The vastance Of the E- distribution is greater than 4 but tends to 4 as n>. As veo then e- distribution become novmal . 5. The vasianee = v_ , if v>2 and Ka >! alunys - v-2 APPLICATIONS (OR) USES OF |- DISTRIBUTION 4.70 test the signiticame ot a single mean. 2. To test the aignificance auijjwunce betwen the two 4ample means . 3. To beat the table value of b, then veject Yo PROBLEMS BASED ON _t- DISTRIBUTION : @ Ten oil tins ave taken ftom an automatic pitting machine. The mean wetgnt of the tin is 15:9 Kg and ie. 3 Cright tailed) Caleulated value of t: Test statishes t. Rf. 169 - M62 ay sNin=1 1¥-2) Gy = Table value of t for v = n-t=21 , degree of freedom at s-/- level of aignificance = L¥2 (one tatled) At 5+ tevel, calculated vatue of t > table value oft i. Reject Ho The. advertising campaign is successful. OA Apave past manufacturer 1s making «spare parts with axle diametex of 0.700 inch. A vandom sample of 10 parls shows a méan dlamelue of p-7y2 Inch with a sD. of Dou Inch . vanity whether the work tabulated value of t + Reject Ho The preduct is not meeting the «pecitizations ® From ao lange population of unemployed youths , a vandom sample of 25 ix table value of t v+ Reject Ho : The diffevene between the population mean ard sample Mean are aigniticant é 4S 4 confidence limite for mean fs % t boos x FAS t 2.931% * = 4s +1599 2 WS -1-5493, yisttsaa) = (39.402, 43-093) 94+ econfidene mits tor mean Ho = Xt boo X—4 n= Wes + sayy x3 Ib = Ws t 22) (urs 2.21, Urs + 2-21) = (34.29, y3.41) © The nine items of ample had the Following values : us, 44, 5O, 52, 4%, 4%, 44,53 and si . Dow the mean of the nine items attter aigniticantly from one ascumed population mean of yes? Solution: Given: = 4, fle ues = a : [ast gat SO+S2+494 4F4 4A4 53 +51] = ANE Gutta q To ting : Sample $.D. 2 Method 4: S22 Sxd_ (2) = 22 _ oy 7 7 7 = [us?+ ux% 507+ sa 4 49% UF% ugh sa°45I7] = 4a. = 2419 - 2uIl. @001 = 6.09aq Az 24bQg Method 2. [x | @-%) | us| -4l 1b-8F Qt] -2-1 aus so} 0-34 0-79 s2] 2-99 B3S NB) oh ngs = (HRY = sy.87 yt] -20 yus 4a} -on 0-01 53| 3.99 1513 Si} 4 .9q 3:sF WKT, $22 Ze)" saa¥ = 6-09bbF nv -£ = (e0teey = 2-4bar (ov) Wet S = SUew*_ SHY E-geats nt (ie Aloo 4°: (n-1)S? - 2X 695345 _ Loqusy n “. Ae Vooteet = 2-ubat Null Hypothesis (Ho): There is no significance diHlerence blw the population mean and sample mean (PE NES inches . Altemative Hypothesis CHD: Be US inches Calculated value of z Test statishies ¢> Ko M 2 AVN FS ygycy + shina 2-4 bal TF “TEL = Bye leva of aignt ficonee = 9.31 At s 4 level, calculated value of t z table value of t . Accept Ho The difference of means is not sigrifieant Table value of t for v=nt~9 . degree of freedom at si 4. The wages of 10 workers token iat a trom-a factory ove given ae: wlages : 599, 592, S70, S49, 5¥2, SYR, Sx0, 572,596; 594 Ts it possible that the mean unge be all workers of this factory cold be R590. aw Solution: eae Given; 210, P=s?0, » aL bres: Sere SHO 4568 + S424 S¥ETSIOT 672 1 + S464 sey] + SHO. oH sere eevee eee se gels eee eer ee To tind Sample sD (a er oe e Methad 4: fea wg a a Sat sar 2 - &3) ar 7 a th 7 gee as [ove S38 4 sto% 563% s32"+ 543" S¥0's 57245404 Sal ]-(#) S's S3184Ib _ (ste): 10 Ss B3IBYL- 33176 Bre 66-6 u..” eee tar . Bz 9.099388 : See METHOD = 205, 2% net eee ee os | he | a-O*}, , SFB], 2h hey sr] 4 he sto| -b Bb ser | -8 by we Bes (nO bt = 656 ; gee : ae . be Jos b = 809938 Null Hypothesis (Ho): No signifi cance diltitence beboeen. the population mean and sample mean’.[= 580. i ‘ Altasenative Hypothesis a tae | Galeulaled. volte of + 7 Feat alabishes | te Bek, _SH°590_ = 741542 ein 409q3a3/1q |, . tl= b4@1sq2, Table yalue of t for ven-l=4, fee of halon ot st eve of significance = 9.26 . At sy» level, cabewlatid value of te. table. value of be w. Aceept Ho ~~ eee Thus the wuan wage of all» woorkune of the factory could be $30". : ; 3. Sandal powder i< packed into pacleeds ‘oy a mathivie A random somple of (2 packels is drawn and their uoerghte ave found. to ber Cin.eg-) 0-49 OFU8,.O-KE, 0-49 :,-0-44. 0-50, OST, 0-4, 04S, O-F0, 0-51, 0-49: Test it the avinage packing con be taken os Kg. : sation: : ak Jee Dasssoaeke To tind: Sample SDM ssa i eee tee . Bee ee See ae el fosetr O1Ug tH ONE + ONS ONG 10:50" O51 aster $0 -uQ%+ O.YO™+ O'S074 O-SI™+ 0-49? are 298% _ (yay 2 4S = 0.24095 —0-2u0! S62 'O0008' °° Pear Here ‘ + = 0-012 2444S i Nult Hypothesis tHe): Thue is no aigniti cane digjnens behueen the populabionG «awyle mrdin uo Albumative Hy pothesis Cu): PF OS . a y Coloutated value of t test statioties = Hp. 004-95 2 2 ay gog Alf n-\ o01224t]in 2. Ie 2-F08DI3 Table value of & for ven-) =i dey vee of freedom at s-I- level of signifi cance 22-20. at st level, calowlated value of ¢ > table value of + ». Regect Yo Average packing cannot be taken to be Or Ke @. The average breakin: strength of the steel vods is “pengied to be 1¥.5 (in units of 1000 %g)- To test this, sample of 14 rods were tested and gave the telowing vesulls: 15, 18, 16, 21, 19, oi, 14, 1F, 1S, IF, 20, 1%, IF, 18, Ls the result of the expsaiment aignificant ? Also obtain the 95 7% confidence interval tor the avenge breaking strength : Solution: biven*s n-th, = IFS lin units of 1000 «g) Zs sa, L[strarivtatiq¢ arr ize lgristlyt 20r arta] n- WW = 280 , 14. a7 yo To tind: Sample S:D. 6 tat - Ge)". Se" _ aot vis zt IS 194 1B tor tty OIF HIF 1g? FISH 4 toot 19? 192+ .19% sia a ~ (4. as Hoy 2°2 322, y2@sF| - 31B.84F520 = 555105) 4=L3S4Yy 233 Null Hype thesis (Ho): There is np signiticance ddigiwerne between the populahion and sample muan. ps 'tS Alternative Hypotinesi CHD: PFE. colowlatd value of tb. Test statistics t: _A-K | WEBS HYD ~ 14-5 41fnr 19244233 /Yjs + b= 06933352 Table value of t for v= n-4t=13 degree of freedom at $1 level of significance = 2-16. At 5+ jevel, calewlated volue of t 2 table value of L. -) Aceepk Ho “The difference of means is not significant as), eonfidance Limite of = K+ boos X n-t Se aay x L8Phy 288 13 = 14-5 70-3130 42 = (4-3 + o-s3002 , 5S -0-313042) = (4913042, 1196058) a “¢) TEST FOR DIFFERENCE. OF MEANS . i {itt ae Et Considit two independent Small nitna-2- WORKING. RULE: ‘Write the sample size oy and na, means H,and %', tandand deviation s and so i) State nuw hypothesis Ho and albunabive Hypo treats Hy Wi) Find the cateulabed value of by using test statist. te U1 -% whine Ss msi=+ nese? and > ning -2 SJnth Hi COnsiden ttl tf it becomes nugative Ww) Wat able value of +t for.vsnrm-2, degrees of Preedom at s+ level of significance | : ee ¥) Condtusion: ia i e ‘ Ma 34 the. calculated value.of t < table value of t, acuept He) or. “WY caleutated vawe of t » table value of t, vejeck Ho . PROBLEMS: ©} Samples of two typus of electric bulbs were tested tor length ot whe ard. the fetlowing + T Type _[Sarele size] Sanple Mean} sD pa 3 1234 hours | 36 hours I # (036 hours 4D hous Ts the diflurence in the mean sudpieturt to waxxant that hype I is supsnion to type I Megavding the length of Lite? Souution : Given m=4, OF , TH = 1234 , 2 =1036, 6=36, 62+4O Null Hypothesis (Ho): Two types ot bulbs have same ute time .tord Wise is no elanifieance le pycunce blu sample meavr ve, =p ANucnative Hypotlwsis (Hi: fu >Hx (Type £ ig supinior thon Type Coleulated value t: Test cbatishs t- _%1-% ee sfare wher 2. ms? 2h =p fe bos mst ene | BSber ho esq osiq. mith -2 att-2 - 8 =VibS4.0F64 = WO.F3 1% Heme £ = _!2 30-1036 worsle frat - ate 4.39243 Table value of t for v= mtm-2 = 13 degrees of tveedom for one tatled at sy level of significance =1:77. At sv level, caleutated value of t > table value oft, Acupt Ho Thus type I is autinitely superstar thon type I Int luigene bust of two groups of boys and girls gave the folowing sHesults » Group of 12 girlk: Mean = 94 , $.D.<10 voup of 3 boys: Mean =91 + Sp=n. Examine if the arfloence of means ave aigntteant . Solution: Given: my=t2, no=B, DY = WY, Me =91 , oat, Kaen. Null Hypothesis (Ho): No aignitreance ditlwene blw 2 sample Mean. fe, bi spo Aluanative Hypotnisis CHI: py + We Caleuldted value of t: Test statishes t- 1-2 _ Slant my Where S*= Pisi®+ mss" oto ty 3:12" 30.4% Mita -2 1249-2 S= Iyd Hee b= _Su-ei _ ogee 43 fo nua AS a Table value of t for v= air. -2=19 degree of treedom at S+ level of Aigniticanee = 210 At s+ level , calutlated value of . Aeetpt Ho The difflaence of mean is not significant t ¢toble value of b. ®] two independant samples f sine 16 ave taken from novmal populations . The sample Means are 26 and to and. vanltanves ave t and y nespectively. Ts the difference betuuen the means significant 4 Solution: Gtiyen: p= DoS , % =26, Xa clo, S¥, Ss 4 Null Hypothesis (He): Thane is no significance diffemence blw the 2 sample means .i-e., Wy) =po Altusanative Hypothesis (Hi: pre pa Calculated value of t: Test slatisties t= _% -%2 ieee Slatin white $?= mSi*t ase? | IF + ISH 5. gq2au nitne-2 IStiS -2 $= 2WU2WtS Hence t = 26-10 = 12-0506. a. a wars AE cats) Table value of & fov y= Mtm-2-0¢. degrees of freedom at S+ level of aignificane = 2-05 At S+ level conetusion , calawlated + > table t + Reject Ho Ditlanence of mean is aignit reant The means of two Kamples and 143-32 of sizes 4 and + ave I4b-yo respectively. The sum of squares of the deviations prom the mean ave 24.q4 and 19-43 reapectivaly. fon the samples be tonsidined to have tome from the normal populations of equal means. Solution: oiven: meq, Maes, FI9b-YD , 1B = 19.92, Sla-Xe)*= 2b-4y, = big -He)*= 12-73 Ste 2 eu R) > Sl%2-K2)* ob-ay 4 9-73 eae tee eee eae nitng -2 art -2 = 39-2621 - $= 1gooel Nult Hypothesis (Ho): there is no 4ignificare ditfvere blw the two éample Mean ie, Mi=pr Alwmnative Hypothesis ¢Hid: flr # Ho Caleulated value of t: Ce Saas co El BC mae a ice : fx s hrs ree far L calculated value of |tl= 2-b34g. Table value of t for v= mtm-2 =14 , degree of freedom * at sx level of significance = 21US At St tevel, eqleulated value of t > tabulated value of t + Reject Ho The ditlerene of mean is agnificank . The avanage — urmbut of axbitles produced by two machines pt day ave 200 and 250 with Table value of & for v= Mitme-2=48 , degree of freedom at iy \evel of signifreance = 2.59 At $+ level, eatculated value of t > table value of t. w Reject Ho Thus both machirus ave not equally efficent . NOTE: In the above problem , the sample sizes ave equal . that is my =M =25 = nC aay) . Then tore eaponding test «statushe| is te Hp Le 200 -250 = -%.65049 S174 Ss? 2074052 Vv n= Vo as- The following table gives the value of protein from Kangeyam cow's mill and buffalos mitk By Using “t’ best examine if these diffarenses ave significant. couls milk. 140 14S 200 202 1ea5 125 Buppalors milk : 2:12 2.00 220 2-4Us 2.20 2.10 Soution: Cet ou and a, be the values of protein from kangeyam cows milk and buffalo’ milk . Given: Ay=2b6, Mr=b. Hs Soy. WS? 2192 o 6 a= ~ BOF. seg Sas ne =] my Xe (m-ni) | Cx 50F | (t-29 | Cava rq aa -0-02 0-0004 | -0.053 | 0.003364 45, | 0 0:03 | 0.0009 |-0.172 |o.o31684 2 2:2 [0:02 |0.00b4 |0-022 |o.c00ue4 2.02 | 2.us Ol 0-01 0.242 |0.043434 as 2.0 -O-0F |0-004% | Gor9 |o.c004ay Las | a “0.0% | 9°00U% | -0.073 [o.00K084 O:02%5 O60 34 | Se BOUT) + F bad? ODES + ON EOAW _ H.o1y 554 SSA SME eer eRaieke De 0 § = ni+ne -2 btb-2 $= OQa25 Nutt Hypothesis (Ho): There is no aignifieance di Henene Uw the two sample means fey Mie Ma alttnative Hypothesis CH): fy + Ha Colouated value of t bs Ha iieve |. | _béo soe 1 3.4993 $7, . AL Tit Oll4395 Ste + lel= 3-42.93 Table value of t for y= mt O-2 =10 , degrees of Creedom at s+ tevel of significance = 2.33 « At sv levd, calculated vawe of t > table wlue of & 2 Refeck Ho : The diflazence of means is significant : ~ the fellowing frqures vefen to observation in 2 indeperdunt aowples - : Ee : Somple rT : 25 380 29 84 24 20.13 32 22 3g Semple I + 4O 3y 22' § Sf 40 30.23 34 'I¢ Avalyse whethex the samples have been dvawn from the populations of equal means + Sotution: : let ou and %2 be the values of Sample r and sample I vespectively, Given: M=10,n2=10; Ti = Sey _ 2bb He = SH. 28) ogy 10 No [a [ee [em | coat | Cm | ee i 25 40 whe 2-56 Wg IUbbt 30 | 34 aq | Sb} 6.4 Bua 28 22 by 9b ~bl 3Hal 34 8 ty Su#6 | -20) | 4O4-O1 ay 3 ob bb 2.4 aut 20 4o -bb usse | 0g (Wher 1B 30 13.6 | igy.ag fo 1% 3-61 32 23 sy 24:16 -s4 26-01 22.) sb | Wb! ate Pel 62.41 32 Vt Wey 124-96. | -IEt- 123-24 266 231 J 4 Bb. Y : 932-9 S22 3 um) 4 E-Pub ut aang = 91-6999 nitne-2 !O+HO -2 S=4.030@ Null Hypothesis (Ho): There is no atgnitieane diffrence blw the two sample Means ‘.e., Mi=pe2 Albxnative Hypothesis CH: [ly Mo. Coleulated value of E: Test statistics t= _Si- % Slat m wi ltl = 0-3F124. = ~O-SFI2Yy Table value of t forv=nitn.-2-19 degree of feedomat sr level of significance = 2.10 At s+ level, calculated value of + < table value oft . - Aceept Ho ‘Thus the amples have been dvawn from the population of equal means. In cevtain e%patiment to eompave two types of pig, foods A and ® , the following results Of increase in weignls were Obstaved in pigs - Pig No YL 2usB gees & 8 Tetel Inurease in Food A : 49 53 S51 52 4% SO 52 53 uot wh. in Ib: FOoD @: 52 SS S2 53 5p 54 54 5B 423 Assuming that the two samples of pigs ave Independint, can we indude that Food B Is bettin than Food At Solution: let 9 and a2 be the values of Food A & Food ® respectively, tiven + m3, M223; H 2 2%, YOR Lsoses ; %- B%2, 423 -so.ees m 3 Mn 3 a He | em | teemer| Cwm) |ta-e)* 44 52 -1.9%5 [3.515625 | -0.975 jo.tb5625 53 55 aioe |4-s15625| ot25 [4515625 st 52 plas |o-01s625 | -D-375 [0-%5b25 52 BS blog | 1265625 | o105 0-015b25 ue 50 3.975 |!8.015625 |-2.9F5 [3.065695 50 Sy -0.375 |0-%65625 | 1-125 [2056 25 52 54 tios |iaese25 | te9 | .obseos 53. 53. 2125 | 4.515625) 0-25 |oolsees yor 423 30-375 16.995, > 2 ss Z Gur dil ys len Wa)", 30-995 16875 swore nite -2 849-2 o. SFt-Bubs Null Hypothesis (Ho): Twte % nD algniteance diferente blu the two sample means .Le., fli=po Aluanative Hypothesia Hid: pr epee. Caleulated value ob: 50945-5975. _' oy igeg Test statistiés t: _ % ae : i q — j 6 fark 18ybg x + 7 co ltle 2.1654 Table value of t for V=nitn2-2=1y degree of freedom at s/- revel of aigniticance =1-¥b . ee At 54 level , caleulated value of & > table value of t-. ws Regeet Ho te ‘Thus food B. is bettir. than food. A-. A group of 10 vats fed on diet A and anotnn gvoup vf 3 vale fed on dieb .B veeprdid the following: ineuase in woetght (gma) | Died PA ities terra tere ieeeee me eee Eee Het EEE) | Det B: 2 S 6 8 10 t 2°28 \ Does is shows suputiority of diet A over, diek & 7 Solution: let a) and x2 be values of Diet A and Diet @: respectively. Given > miziD, N= Bs ee Se, BB yy; Me Zoe Ws nm 8 21 i | oem | Com | (om) | Oe mD* Ss 2 ay 1% 3 4q 6 s OY O-tb aos 4 g 6b be 2.56 t \ ' 3 “5-4 24.16 3 a 12 10 Sb 31.36 5 as 1 -2.y 5.46 “4 16 3 2 -3-4 1. 56 3 di 2-6 & #6 3 4 6 “Oy Outb 10 db 12-% by 40 102-4 32 Se = (uur te Ge-GF 102-0782 sos nite -2 5 1048-2 +. $= 3.39485 Nutl Hypothesis CHO): There is no significance dittxwence blu the two sample means je, hi= He Alwanative Hypothesis CM): fa > fa Caleutated value of E: Test stabishes t= _5i-% . __b4-S | o.acasag 5 34d. she asauss| te Table value of t for v= ntne-22tb degree of treedom at s. level of significance = (45. At s+ tevel, calculated value of t < tabulated value of t - Aceept Ho [pitloreme is not aignthicant ] We cannot eonchide that Dret A ia Aupevior to Dret 8. 29 " Achool boys were given atest in statistics. They were given, @ months tution and a test was conducted at the end of it . Do the mavks give evidence that the students have benefited by the extra coaching ? Boys cea 2 3 4 5 6 F BAwW UH Marks in Ttest : 23 20 14 at 12 20 19 IF 23 le 1q Marks in atest ; 24 1 22 19 20 22 20 20 23 20 19 Solution: Since the data ave correlated. 1.2.,Marks obtain b the Nl Atudents in © best (before coaching) and in the © test (aftev coaching) are correlated. let ai and yj; be the marks obtained by the boys before and after coaching. We will use t test for single mean of the Paived data. tet d= wy , then d =a-g (on zd. a 4 |dsa-y | Ga @-a* 23 24 = 0.09041 0.00826 20 a 1 2.0902! 4.3419, 1Q 22 s A-4090F 3.644 b2 2 12 3 44-0909) (6.73554 1g 20 2 0.90409 | 0-82644 20 22 -2 -0.20909 | 0.22644 1g 20 =a -0-90909 | 0.8264U4 1% 20 “3 -1.90904 | 3.64b02 23, 23 ° 1.09041 1.19008 16 20 “4 ~2.90909| 8.422 1a 13 u 2.09021 4 SFF “2 4Yy.90904 @ a= Za . -12 7 I $?= 2G@dY_ yy-a0doy 4490904 n-1 Het 1. &FQMNUES Null Hypothesis (Ho): There is no significarne difference blur the Marks obtained by the same set of students Alwamative Hypothesis (Hi): Ditlenence is significant . Test slatistre t-_d . _-2I0 2-1. 40433 svi QUES) . Ith = 1-40%33 Table value of t for v= -1=10 degrees of freedom at 5/- level of aignificarne = 2.93. At sy level , calculated value of t 2 table value oft. *. Accept Ho Ditteverce is not signitieant . ie, The students have not benefited by ¢oa.ching . Memory capacity of 4 $2) a The degrees of freedom of this statistics is ve(m-l,m2-2) APPLICATIONS (or) USES OF F-TEST It is used to test whether two independent samples hove been dvawn from the nermal populations vith same vaviances o> (ory W It is used to test whether the bwo independent estimates of the population variance are homogenous ov not. WORKING RULE: write the samples size mand Me , variances 62 and $7 li) Stake null hypothesis (Hc) and altexnative hypothe sis(H) ii) Find the calewated value of F by using test tatiatre 2 fy ae (et 45): Sr w) Write table value of F for v= ort, m-2) degrees ot freedom at sx tevel of aignificarne. v) Con clusion : U) TF caleutated value of F < table value of F , ateept Ho Cor) WW) If catculated value of F > table value ofF, reyect Ho PROBLEMS: © Two yandom samples gave the following results : Sample] Sixe |Mean| sum 0! squares of deviations from the man ba to 1s qo Tr m | oy (09 Test whether the difference between the variances iS aignittcank at sy 2 Solution: Given : M=10, Maria, = (-2)F = Qo , = la-maY = 108 Ss Ele 7 = (as -®a 108. a.91¢2 oe n2- 33 r Null Hypothesis Lo) on? = o® Altevnative Hypothesis CH: oh? & oy? calculated value of F: Since $1? > $2") then F- test becomes iene = elo: i = = bOl8s2 5° 4919.2 Table value of F for (mit, 92-1) = (4, degrees of treedom at sy level of aignttreance = 2.90. At sy tevel , calenlated value of F < table value of F vs Aceepk Ho -. The vavianues of samples are not signitveant . @ In a sample of 3 obsewations , the cum of the squares} of deviation of the sample values from the sample mean was W-S and in another was found to be 101.4 . variames ts significant. Aample of lo observations ,it Test, whether the diffevene of Solution: Given: m=, Da=10, = (a-R)= W-S , S (X2- Kad 101-4 B= Ztu-T)?_ qu niet re * t $.° = 2 (w2-GF_ ont Ma-l ig ‘Nutt “Hypothesis (Ho): aj? oa Alternative Hypothesis CH): o> 4 on% ° Caleulatedl vatue of Fs eee Since Si? >S.%, the: F best becomes . F: SP ges eed 7 roe PS8FA Table value of F tor (nr-t, ng=l) = (4/10) degrees of 'treedom *: at s+ level of significance. = 314, eee Heer EEE nacre mene ee eee ct At 5+ level, eae value ol: Fé table value of F st “. Aceept Ho “The variances ‘ot samples ave nok vigil feant . woe tat der be ane i : ® ™m a sample of: iw.-observations the sum of ‘squares: o+ deviations of. thes sample values from the’ sample mean was 120 and in: another sample of--12- observations jit was, found to be 3ty, Test,. whether the signi fieance ob-vavianees~ is significant at 57 level 2 Solution: Given: m210, M2=l2, & Ou-%) 120, 5 (ta-&a)*= B14. 2 SP = Sc te 120 = 13.338 sees (Pe orl Sp = Elat2-%2)* | Mo-t - a Null Hy pothesis (Ho): 27? = o3* Alternative Hypothesis CH: <1? 4 3” calewlated value of F: Sine $7 > 5:7, then F test becomes Table value ofF for (mint,na-) = (1,9) degrees of freedom at sy level of signifi cance =a. At sy level, calculated value of F < table value of F » Accept Ho The variances of sample are not signiticant . @ Two rovdom «amples of 11 and 4 items show the standard deviations of their weights as 0-8 and 0-5 respectively. Assuming that the weight distributions are normal. Test the hype thesis that the true varlances ave equal, against the alternative hypothesis that they are nol. Solution Given: moi, M2=9 4 $508, S2=0'5 eae fuse = wlost _ or0y -e 10 = 2 2 2. So = Mase” 2 4.0.5) _ o.gins Me-! 3 Null Hy pothesis (Ho): oj? = og? Alternative Hypothesis (Hi): ej + oa ee | calculated value of F : Since Si? > $1 , then F- test be comes Fe O-F0y 0.29125 rd = 2.503) Table value of F for (n-1, n2-1) = (10,3) degrees of freedom at 5» level of significance = 3:35. At sv level, calculated value oF ¢ tabulated value of F . Adupt Ho -. Variances oF sample ave not significant - © Wwo independent ampes of sizes 6 and 5 from a normal population had the following values of the variables Sample I ro 0B NE Stay sample Z: le 14 13 te 19 De the estimates of the population variances ditter signihicantly at er level? soluHion: let o% ond %2 be the values of sample I and sample Given: 2S, Da=b; cya ieee = Bu. e Ho = 2% = MW iy x eee s @e Te onl a Ae eee oah | (aca) | e220" 13 Io Ib OF Ole 13 4 l 2b bt 2 | 12 4 34 | sb 's 16 t 1 Ou ote 2 13 “2 4 1b 9.56 Vy ° ° sy g2 26 212 = Cw-u)*= 26 and s(x-me) = 212. Se Eee Ki 2b _ gn ment i _ Sane = lata -52)* 24 _ 5.5 n2-t s q Null Hypothesis (Ho): 71? = og? Alternative Hypothesis CH): 7? 4 es? Gleulated value of F: since Si? $37, then F test be wmes 2 Fe $2). $3 ~)01403. $2? 2 a Table value of F for (m2-1, m-I) = (4,5) degrees of freedom at si level of significance ~ S19. At sy evel , calewated value of F < table value ob F ‘. Aceept Ho Samples came from populations of equal vaviance . @® Two \andom «amples drawn from sizs ¢ and # had following values ot the variables, Sample A 9 fie k3)- Ut 7 jie ty Sample B: 10 12 10 14 4 B 10 ‘Do the estimate of population variame differ aigniPreantly - Solution Let 2% and % be the values of sample IL and sample IL respectively. Given: m=S, M2=b. oy | 2 | e- | e-my | (M-%2) amy] 9 | 10 |-2.45 |4.5025]-0-4296]0-193% i) (2 -O-FS | 0.5625] 1.5714 2.U63% 3° | to | tas | 1-562 ]-0-4286]o183% Uo} iy | -0-45 }o.sb25] 3.5714 fi2%suq is | q 3.25 110.5625] -1-4286 | 2.0409 4 @ | -245 [4.5625 |-2.u286/5.9981 12 | tO | 0.25 | 0.5625 |-0-y2 26] 0.193% 14 2.25 |5.0625 ay | 43 33'S 23.9143 He By Ws es 5 Mer Sas L 32. oueee 3 : no EHP = 33-5 and = (ua-RE)*= 23-7143 $= Su -_ 33-5 y gsss! ni-t 7 2 $a = W-FIU9 _ a.qsoa¢ 6 Null Hypothesis (Ho). 2° = os Alter native. Hypothes Is CHI): 09? a5? Since $2 > $2 , the caleulated value of F-test poeormes i “F357! Fe ge = ee = 12108 os 3.95239 Table value of F for (ni-1, 02-1) =C46) degrees of Freedom at sy level of signi Freante 24-21 oy At Ss? level culated value of F< table: value of F Accept Ho : : fe Samptes came from the population ot equal Hanantes | oe a i | @ Two rardom Zs ‘dain from’ two normal populations have _ the values ag, below: HEE Sample, T° 2.19 re [18 99° 20 98 14 oy 26 ‘ ; Sample) ir: 28 32 4O 3% 20 as yo on wt us 30 36 Obtain the estimate ‘of the ‘Variance ‘ot the population land. test whe they the ta populakions have same “Variances 2: Sdution:. Let ‘ain ms be the values of sample x and. sample T vespectively' Given that: M's 4, mete. B= Zor. 194 ol ssh ) Gee Bae y22 cac.yuez. - my a ‘ m7 Ie [ou [ae [ami | -aid] wie [tame 19) 28° | -2.556) 6-5331 |-F-teb¥ [51.3016 te [32 [-us56 foo yse|-B.lobF | toro2a fs, 1h | 4d |-5-556 Ia, gray 4.9333 (23, 3408 | ag | 3e | uN 1-333 | 3:30! ‘ o-uuy [AS251 | rite lay aye 23 fas | 4uy fOl44! [ores foots Prq | 40 l-o.ssb | 2-085!-| 43235 | 93-3008 Ei Al ap” b-S33! |-4.-1LeF |5).3616 : Ee wt a _ 5-47.31 }5.33323 | 34.0274 : “ys RY Ne yan’ | 93333 lau. vane poare 30°] 7 ~S:1ObF lap aqua]. 3b 03323 |p. bauy hay fier Jo: 134-9214 2YF- 666 4) S ElousaF = 184.2219 and = be-X2)*= 3y%..b008 " IbttF ' + = (% Ri) gut -beee. nat a " Null Hypothesis (Ho): ae . ot Altwerative oo td: ge = ‘BI. bob! mr Since 82" > 5/2, the cdleulated F- testers | = 31-606) : peat oe ‘ = ee = 1}. 92421 i Si” Taare nM Table vatue of F -Cna-t,-ni-i) = (1,9) degree of freedom at Sy level of signifi eainée = 3.32, At 5+ tevel, calculated value of F €'table value of F : os Aeeept Ho +, Samples came’ fvom -populakion:: of equal variances . @. Two vandom samples gave’ the Following .dato: Sample size Size Mean Variance 4 "4 ab 12 2 " 1e's os) | Com we eonclude that , two samples can be drawn trom the 4ame novrhal’ population’? Hee Solution : . tee A “norenal ‘population “has two parametins mean pp and vavianee o”. It we test the independent samples have been. dvawn | tom - the sayngle normal ‘population , then the tollousirg L— tests ave nuded. W) bo test the population means are equal Cusing t-test for the diffevence ot means) Wi) to test the varlances ave equal Cusiing F. test toy the difteverces of variances). Before we use t-test, populations should have same vaviantes . Thus we «stant F- test bivst. To apply F-test: Given: m=3, Mest, Sra l2 , g 029.5 2 Sie msi? S202 pa71y3 0; S*s Bs Null Hypo thesis (Wo) : o%= 23% Allusmative Hypothesis (Hi) 1 a” $ o5* Since So > SP, the caleulated F-test becomes, Fe S. 295 _ 9.0062 Si? U3FiGs Table value of F for (M-!, min!) = Uo, ¥) degree of freedom at sy evel ot significance = 3by. At s+ level, calculated value of F < table value of F Accept Ho Somples came from the populations of equal variance [Nove : x4 F-test toiled , then t-test should not be used ] % To apply t- test: Givn: mss, N2=N, Go, MW= 1S, S32 , 4022-5 Null Hypothesis (Ho): nto signiticame diflerene between 2 sample means tes, fi =H2 Altuandtive Hypo thesis CH): Me BE Calowlated value of t= Test statistic t= _M-% _ w 3H “4 3h 2 2 2 Where $°= msp+ mo® - GUr2) + (2-5) _ gigoas Noth -2 19-2 vo SE LUFT Hence E+ _ b= jb-5 _ -J0-081q4 : [ap lurtag 3 + 1 - lel 10-05146 - Table value of t for v=mitn:-2 = 1%, degrees of freedom at sy. level of significance +911) At 5+ level, calculated value of £ Stable value of § + Reyeek Ho The difference of mean is significant 5. The two samples have not been dvawn from the same novmal — poputations. @. A normal population has two parameters mean pe and vastance o*. So use apply t- best for equality of means and F-test for equality of variances « z Sum of squares of deviatiors| Semple | Sixt | Mean) oe he meant ! 12 1y 109 2 10 15 0) Sdution : A normal population has two parameters mean 2 and. veviare o* 0 we apply t- test for equality of means and. F- best tov equality of variances to apply F-test: Given: msi, M=10, stb, Re HiS, S&S (a-m)* = 108, ECoa-%2)*= 90 2, ot SPs = Ou- a) lo s Ss 94-9192 nt ue 3,72 = boa- Gay a0 Ne-l sete ot Null Hype thes is (Hod: 2 = a5? Alnative Hypo thesig oh): a7? 45% Since $i Slate Ma) AL 2 ies Mo-l 8 oe Null Hypothesis (Ho)! 27% = at ” Attetnative Hypothesis Cr Since $17 > $2” » the ¢ealeulated F-test beeomes Fe $* _ 20 = 22 2k 2 ines 4021 Table value of F for (nimi, na-() = (9,9) degrees of Freecom at 5% level of significame = 3.33 me At 5+ jevel, calculated value of F « table value of F° w. Accept Ho Thus, the samples came from the: populations of equal variance Furthuy , we apply E test for their means. Bub, in the given Problem the means ave not given. So we cannot apply t-test for this problem. @. the nicotine cont ent given below Sample I: 21 2y sample IZ : 22 oF can you say that the population ? Satution: let % and x2 be the values of sampler and sample 1. into bvo samples of tobacco are 25 2 3% 22 30 31 36 two samples came from the ae etl = Petes r | ae | (m-R)_ | mi | Cx-mad | Cx) ol 22 “3b (2-9 -7 44 oy | ov | Ob 0-36 -2 4 o5 | 2 Oy Ok et 1 at | 30 | wy 146 ! 1 oe] a | 24 5-46 2 4 3b ud 123 | #4 212 108 2 = Go-M) earn To apy F- test : and = (%2-Xe)” +103 ste Elu-mt _ 5.5 m1 F : Null Hypothesis (Hp): 7122 <3 Altundtive Hypotinesis (Hi): 1? #3™ at Sine %* > SP, the calculated Ftest be tomes fp. SE, Bb .worss : te Table value of F for (m-1, mnt) > (s,4) degrees of freedom at St level of signifi cance =b.26 AL s+ level , caleutated value of F < table value of F. *, Accept Ho . Samples came trom the populakion of equal variances . To apply E- best Se uw) Z toa-mP _ an24108 . y.956 Mit M9 -2 S+b-2 . Sz 3884 Null Hypothesis Cte): No signitieane difference bho two sample Mean . i.e. Wie M2 Alternative Hypothesis a: pe pe Calculated vatue of t: Eel eee eth aai yo Agivg bet , Sait s¥a94 [Eat “MELE LN, Table value of & for y= mtm-2= 4 degrees of freedom at sy jevd of aigni Hieance 22-26. 8 4g” At sr tevd; calculated! ame Obt Te table? value ofE\: > vs Aceepe Ho i The two samples: nave been-dvawn from amé population’ > Thus the amples have been davawn from same normal. i. poputation . welt ead : ye ‘ ie pace seiae ee es fee ‘ A bel weed, wedi, be et 4 2 ioant 3h a | i rites ty % en : aed * each : may4g 4 pint 4 a ’ tek CHI-SQUARE TEST OF GOODNESS OF FIT. Tt O; and & Ci+1,2,... n) be the set of observed lexpurimnental) and expected (theoretical) frequencies then x? is defined as eek [s 2 Ei ot Ej In general, the umber ot degrees Of freedom v is the tolal Tumber of chservations less than the umber ob independent constvainsts imposed on the chservaHons. Tho v= -K. where K is the oumbs« of independent constraints in a set of dota of a co«ervations. Nore : \. The degrees of freedom for fitting a Binomial distribution WB n-l, 2. The degrees otf fveedom for Tilting @ Poisson distribution i neo. 3. The degrees a freedom for fitting a Movmal distribution, is nd. 4. Ip x0 ,then obsewed and expectid frequencies coineide 5. If X* dtstvibution , mean= v and vakianee =2v. “PROBLEMS : 1. The fellowing table gives the number of aiveratt acadents that oceummed dusing the seven dai of the week . Find whether the acetdents ave uni forerty distvibuted over the week. ays -of the weer :'Mon Tue wed Thun” Fre. Sat Gan. Number of oéeidents : IH 0 sR. I th IS ads asl ace acta ai ees Solution: Nutt Nypotiesi Ute): The ascents . axe. unetormly distributed over the week. ve Based on aull hypothesis , the expected freuercies ot the accidents on eath of these days ave Se cet Let 0 be “the ob sevved. fepeny. oe & be the enpeded, Frequency, table, becomes. a Sve ole | w-e* men meat rae fate Po Oe | Os take oat ta] oy it Liy2ee 2 | y & jo.2gset |... rae : tay 4 Oibwage ise eee ce iS} oy ! O-0F14S ‘ yoy Oo | oe fk 214296 . eee 3 ~ Caléulated. value of Xe xf 214296 Table value: of xX? for ven-l<5, degrees cl. freedom ak Sv. level of aigniticame =ior. Since , calculated value of x 2 table value of xt. 1. vee *. Accept Ho + 2 ‘ +. The odes ave eee dated over week @. The table given the numbur of alrenapt acudenls that Ocuunted dusming the various days of the week Test uhether the accidents ave uniformly distribubed over week Day of the week : Mon Tue Wed Thu Fri Sat Numbur of accidents : 14 132 ny 18 1h Sdution: Null Hypothesis (Ho): The aeudenks ave uniformly diavibubed over week. Based on null hypothesis , the expectid pequencies of the acudents on each of these days are BL ety let © be the observed frequeny and © be expectd frequerny the table becomes, 3 [ole [wer | OO ") ty 0 oe) 18 | 1y 1b 114286 12] 1y 4 0.285741 i Ob4u286 | -. Caleulated value of x* IS fay ' 0.07143 Ps o©-ey ww o o e 2-1h29b = 21U28b Table value of x*> v= n-1=5 degvees of freedom at sv. level of aignificanee = Lox Calculated value of x*> table value of x* * Reject Ho Table value of x* w Reject Ho The die is biased. @Tn 120 throws of a single die, the following distribution of faus were obtained. Faves ree 2 3 4 5, b Total Frequency : 30 a5 1g 10 22 Is 1so = Compute the statistic you woul use to test whethue the sesulls constituted Heputations of equal probability . Use chi- Squave test. Solution : Nult Hypothesis (Ho): Die is unbtased. The expected frequeney for ¢ath face of | to bare 120 - 20 let © be obsuwed Hequerey and E be expectid frequency , the] table becomes ole |to-er| SF 30] 20 100 5.00 | P| i207) 2 tee ». caleulated value of x*= z(ex) w] 20] 4 0.20 a to} 20 | 100 5.00 = I2.80 22/20] 4 0.25 ws | 20| 25 125 (2-90 Table value of x* for v=n-1=5 degvees of freedom at S/ level of signiticanee = 11.0% Since , cateulated value of x* > Toble value of x* fy Reject Ho Henee , the die is unbtased © the numbur of acuidenls happened for a paxtrenlar chy in the 5 plas of highway ave 12, 5.10, 6, #. Test whether tne atcidente ave uniformly distvibuted on. any day 6oution: Null Hypothesis tHe): The accidenls ave uniformly distributed on a given day .

You might also like