Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

Embed Size (px)

Citation preview

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    1/60

    BUSINESS STATISTICS

    Assoc .Prof. Nguyen Huu Bao

    Department of Mathematics !"U

    BUSINESS STATISTICS Course sy##a$us

    Instructor: Assoc.Prof .Nguyen Huu Bao - Department of Mathematics

    E - mail: NghBao@Wru. eu. !n" phone: #$%#'&&(

    %ffice: Department of Mathematics"W)*%+, ayson Dong a - Hanoi Cooperator: Mmath. Nguyen !an Dac Phone : #$/,,#,,#$

    Prere&uisite: Basic Mathematics s0ills

    Te't$oo(: Essentials of Moern Business 1tatistics 2ith Microsoft E3cel. Anerson

    12eeney an William .

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    2/60

    )aptop* 4ou 2ill 5e e3pecte to use your laptop in class an to complete the

    Home2or0 5y using Microsoft E3cel soft2are. 1ome 6uestions on the mi - term

    an final e3ams 2ill re6uire using 7aptop. Bring your laptop to class e!ery ay.

    Course content: his course 2ill introuce you to all the important 5usiness-

    relate topics in applie statistics in one semester. his semester 2e 2ill learn a5out

    ata ac6uisition an analysis" ta5ular" graphical an numerical methos 2ill use of

    escripti!e statistics" pro5a5ly istri5utions" statistical inference an regression

    analysis. We 2ill use the statistical capa5ilities of Microsoft E3cel to reuce your

    calculations

    Tests*hese 2ill 5e three ests uring the course. he est 2ill 5e in at least '#

    minus an each 2ill 5e score on the 5asic of %## points.

    Home+or(: Home2or0 from the te3t5oo0 2ill 5e assigne an score regularly.

    here 2ill usually 5e home2or0 assignments ue on uesay each 2ee0. 7ate

    home2or0 2ill not 5e accepte. Missing home2or0 2ill 5e score -( points.

    ,ra-ing Stan-ar-s*

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    3/60

    Tests * / 0 122 points 3 /22 points

    Home+or( Assignments / 0 45 points each 3 65 points

    Contri$ution to Team !or( 5 points

    C#ass Atten-ance an- Participation 42 points

    Tota# Possi$#e 722 points

    /82 9 722 points : A ; /42 9 /5< points : B ; 4=2 9 /1< points : C

    472 9 46< points : D ; Be#o+ 472 points : > >ai# 3

    ,ra-ing Stan-ar-s in C#ass Atten-ance an- Participation *

    - Each a5sent time : - , points

    - 8ame late: - ( points - Each missing Home2or0: - ( points - Presentation: 9 , points - 8orrection the e3amples at class: 9 , points

    Course 9 Sche-u#e

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    4/60

    Day Course Content

    +-%# ntrouce the course. 8hapter %: Data an 1tatistics . 8hapter ( : Bar chaan

    Pie chat

    /-%# 8hapter &: Descripti!e statistics. Numerical Measures

    %%-%# eam Wor0 : Practice Microsoft E3cel

    %(-%# 8hapter ;: ntrouction to Pro5a5ility

    1ome e3amples calculating Pro5a5ility of a e!ent A : P

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    5/60

    (%-%# est for Population Mean

    ((-%# 8hapter %(: 7est s6uare metho . 1imple 7inear )egression

    E3cel>s )egression ool

    (,-%# eam Wor0 : Dra2ing the 1imple 7inear )egression 7ine

    ('-%# )e!ie2 for est &

    (+-%# Test / nform the 8lass a5out Attenance an Participation

    1core

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    6/60

    CHAPTE" 1 DATA AND STATISTICS

    1.1 DATA* Data are the facts an the figures " analy?e an summari?e forpresentation an interpretation. All the ata collecte in a particular stuy arereferre as the ata set for the stuy

    E#ements are the entities on 2hich ata are collecte

    A ?aria$#e is characteristic of in interest for elements

    Data sources* Datacan 5e o5taine from e3isting sources or from sur!eys ane3perimental stuies esigne to collect ne2 ata

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    7/60

    1.4 Statistica# Stu-ies* 1ometime the ata neee for a particular application are nota!aila5le through e3isting sources. n these cases" the ata can often 5e o5taine 5yconucting a statistica# stu-y.

    1tatistical stuies can 5e classifie as either e3perimental or o5ser!ational

    1./ Descripti@e Statisticshere are t2o omains in 1tatistical 1tuies: Descripti!e 1tatistics an heoretical

    Most of the statistical information in ne2spapers" maga?ines" company reports another pu5lications consists of ata that are summari?e an presente in a form that is

    easy to reaer to unerstan. 1uch summaries of ata 2hich may 5e ta5ular" graphicalor numerical are referre to as Descripti@e Statistics

    1.7 Statistica# Inference

    The popu#ation an- the samp#e*

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    8/60

    Many situations re6uire information a5out a large group of elements < ini!iuals"companies" !oters" househols" proucts" customersBut" 5ecause of time" cost" an other consierations" ata can 5e collecte from only asmall portion of the group

    he large group of elements in a particular stuy is calle the Popu#ation an thesmaller group is calle the samp#e

    Some e'amp#es in c#ass

    1.5 Statistica# Ana#ysis Using Microsoft E'ce#

    % ntrouce Microsoft E3cel

    ( Basic perations 2ith Microsoft E3cel& Bar chart 2ith Microsoft E3cel

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    9/60

    CHAPTE" 4

    DESC"ITI?E STATISTICS TABU)A" AND ,"APHICA)

    P"ESENTATI%NS

    4.1 Summariing Categorica# Data

    >re&uency Distri$ution

    A >re&uency Distri$ution is a ta5ular summary of ata sho2ing the num5ers ool to 8onstruct a Bar 8hart an Pie 8hart of BranPurchase

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    10/60

    4.4 Summariing uantitati@e Data

    >re&uency Distri$ution

    he three steps necessary to efine the classes a fre6uency istri5ution 2ith6uantitati!e Data are: %. Determine the num5er of non-o!erlapping classes (. Determine the 2ith of each class &. Determine the class limits

    E'amp#e: *sing E3cel>s PFAB7E )EP) to 8onstruct a Cre6uencyDistri5utionA ot place a5o!e the a3is

    Dot p#ot* A hori?ontal a3is sho2s the range for the ata. Each ata !alue isrepresente 5y

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    11/60

    Histogram* his is a common graphical presentation ata 2hich can 5e prepare forata pre!iously summari?e in either a fre6uency" fre6uency or percent fre6uencyistri5ution

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    12/60

    NUME"ICA) MEASU"ES

    /.1 Measures of )ocation

    Meann statistics formulas it is customary to enote the !alue of !aria5le 3 for the firsto5ser!ation 5y 3% " for the secon o5ser!ation 5y 3( . Cor a sample 2ith no5ser!ations" the formulas for the sample mean as follo2s

    ix

    xn

    =

    N8E:%. he mean is the a!erage !alues(. f the Data are for a sample " the mean is enote 5y x

    An if the ata are for the Population" the mean is enote 5y

    Some e'amp#es at c#ass

    Me-ian* he Meian is the !alue in the mile 2hen the ata are arrange inascening orer

    N8E:%. Cor an o num5er of o5ser!ation" the Meian is the mile !alue

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    13/60

    (. Cor an e!en num5er of o5ser!ation " the Meian is the a!erage of the t2o mile!alues

    Some e'amp#es at c#ass

    Mo-e* he moe is the !alue that occurs 2ith greatest fre6uency

    Using E'ce# to compute the Mean; Me-ian an- Mo-e

    Some e'amp#es at C#ass

    /.4 Measures of ?aria$i#ity

    "ange* he simplest of !aria5ility is )ange

    "angeG 7argest !alue 1mallest !alue

    ?ariance he Fariance is a measure of !aria5ility that utili?es all the ata. f the atafor a Population" the a!erage of the s6uare e!iation is calle the PopulationFariance an is enote 5y ( . f the ata for a sample" 2e shall call 1ampling

    Fariance an enote 1(

    :

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    14/60

    (

    (< =ix x

    s

    n

    =

    (

    (< =ix

    N

    = :

    Stan-ar- De@iation*

    Stan-ar- De@iation*

    Population stanar e!iation: G (

    1ample stanar e!iation: s G (s

    Using E'ce# to compute the Samp#e ?ariance an- Samp#e stan-ar- -e@iation:

    Some e'amp#es at C#ass

    Coefficient of ?ariation

    Coefficient of ?ariationtan

    < %##=s dard deviation

    Mean

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    15/60

    /./ Measures of Association $et+een 4 ?aria$#es

    Co@ariance* Cor a sample of si?e n 2ith the o5ser!ation (- Dimension

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    16/60

    Using E'ce# to compute the Co@ariance an- the Corre#ation Coefficient

    Some e'amp#es at c#ass

    CHAPTE" 7

    INT"%DUCTI%N T% P"%BABI)IT

    Pro$a$i#ity is a numerica# measure of #i(e#ihoo- that an e@ent +i## occur

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    17/60

    7.1 E'periments; Counting "u#es

    E'periment is a process that generates 2ell-efine outcomes

    . E'periment E'perimenta# %utcomes

    oss a coin Heat" ail 1elect a part for inspection Defecti!e" Non-efecti!e )oll a ie %"("&";","'

    Samp#e space for an e3periment is the set of all e3perimental outcomes

    E'amp#e

    Counting Techni&ues

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    18/60

    1 Mu#tip#ication "u#e: Assume an operation can 5e escri5e as se6uence of 0steps an

    he num5er of 2ays of completing step % is n%an he num5er of 2ays of completing step ( is n(for each 2ay of completing step %

    " an he num5er of 2ays of completing step & is n&for each 2ay of completing step (" an so forth .he total num5er of 2ays of completing the operation is

    n G n%. n( n04 The num$er of permutation of n ifferent elements is nJ 2here nJ G

    %.(.&.n/ The num$er ofpermutationsof su5sets of r elements selecte a set n ifferentelements is

    J.< %=.< (=...< %=

    < =J

    n

    r

    nP n n n n r

    n r= =

    7 The num$er of com$inations " su5sets of si?e r that can 5e selecte from a set

    of n elements is enoten

    r

    orn

    rC

    J

    J< =J

    n

    r

    r nC

    n r n r

    = =

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    19/60

    E'amp#e* Crom a es0 of ,; cars ra2 & at ranom. Ho2 many ifferent 2aysra2ingK

    E'amp#e* A 5atch of %;# semiconuctors chips is inspecte 5y choosing asample of fi!e chips. Assume %# of the chips o not conform to customer

    re6uirementsa Ho2 many ifferent samples are possi5le K

    5 Ho2 many samples of fi!e contain e3actly one nonconforming chipK

    E&ua##y )i(e#y %utcomes*he Pro5a5ility of an outcome can 5e interprete as oursu5Lecti!e pro5a5ility or -egree of $e#ief" that the outcome 2ill occur. Whene!er asample space consists N possi5le outcomes that are e6ually li0ely" the pro5a5ility ofeach outcome is %N

    7.4 Pro$a$i#ityof a e@ent

    >or a -iscrete samp#e space; the pro$a$i#ity of an e@ent E ; -enote $y PE3 ;

    e&ua#s the sum of pro$a$i#ities of the outcomes in E.

    >or e@ery A in samp#e space; PA3 : m

    n !here m is the num$er of the

    outcomes ha@ing e@ent A to appear an- n is the sum of possi$#e outcomes .

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    20/60

    Assigning Pro$a$i#ity

    Pro$a$i#ity is a num5er that is assigne to each mem5er of a collection of e!entsfrom a ranom e3periment that satisfies the follo2ing properties: % P

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    21/60

    E'amp#e 4* >rom a -es( of 54 car-s -ra+ three at ran-om. >in- the pro$a$i#ity

    that there +i## $e e'act#y one ace among them.

    Solution:Denote the e!ent 2e are intereste in 5y A he num5er of elements of

    iscrete sample space is ,(&C . he num5er of 2ays ra2ing one ace is;

    %C . he

    num5er of 2ays ra2ing ( others cars < is not ace = is;/

    (C . hat means the num5ers

    of 2ays ha!ing A to appear shall 5e; ;/

    % (.C C . 1o " P

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    22/60

    7./ Some Basic "e#ationships of pro$a$i#ity

    A--ition )a+

    Union of t+o e@ents* *nion of t2o e!ents A an B is shae in &-th figure a5o!e" thatmeans A appear or A appear

    Intersection of t+o e@ents* ntersection of A an B is shae in ;-th figure a5o!e" thatmeans A appear an B also appear

    A--ition )a+* < = < = < = < =P A B P A P B P A B = +

    N8E: f A an B are e3clusi!e e!ents then< = < = < =P A B P A P B = +

    Some e'amp#es at c#ass

    BA

    A

    B B

    A

    B

    A

    B

    A

    A

    A B A B A B A B A B Ac G A

    A

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    23/60

    Con-itiona# Pro$a$i#ity

    Cor any gi!en e!ent A an B" the pro$a$i#ity of the e@ent A appeare- +ith thecon-ition e@ent B appearis the ratio:

    < =

    < O =< =

    P A BP A B

    P B

    =

    he pro5a5ility of e!ent B appeare 2ith the conition e!ent A appeare is efine 5ysimilar 2aysSome e'amp#es

    In-epen-ent E@ents* 2o e!ents A an B are inepenent if< O = < =P A B P A= or < O = < =P B A P B=

    Mu#tip#ication )a+

    Multiplication 7a2: < = < = < O =P A B P B P A B =

    r < = < = < O =P A B P A P B A =

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    24/60

    N8E: f A an B are t2o nepenent e!ents then< = < = < =P A B P A P B =

    T%TA) P"%BABI)IT "U)E

    A collection of sets E%" E(" " E0such that E% E( E0G 1 is sai to 5ee'c#usi@e

    Assume E%" E(" " E0 are 0 mutually e3clusi!e an e3hausti!e e!ents . hen

    % % ( (

    %

    < = < O = < = < O = < = ... < O = < =

    < O = < =

    k kk

    i i

    i

    P B P B E P E P B E P E P B E P E

    P B E P E=

    = + + +

    =

    Some e'amp#es at c#ass

    n the case only t2o e!ents A%an A(2e ha!e

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    25/60

    BAESS THE%"EM

    % % % %

    %

    % % ( (

    < = < O = < = < O =

    < O = < = < = < O = < = < O =

    P A P B A P A P B A

    P A B P B P A P B A P A P B A= = +

    Some e'amp#es at c#ass

    Using E'ce# to compute Posterior Pro$a$i#itiesE3ample in page %/$

    CHAPTE" 5

    DISC"EETE P"%BABI)IT DIST"IBUTI%NS

    5.1 "an-om ?aria$#es

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    26/60

    Discrete "an-om ?aria$#es* A ranom Faria5le that may either a finite num5er or aninfinite se6uences of !alues such as %"( "

    EIAMP7E:E'periment "an-om ?aria$#e Possi$#e ?a#ues

    %. 8ontact fi!e Num5er of 8ostumers #.%.(,costumers (. perate a restaurant Num5er of 8ostumers #"%"(".

    Continuous "an-om ?aria$#es* A ranom !aria5le that may assume any numerical!alue in an inter!al or collection of inter!als

    EIAMP7E:E'periment "an-om ?aria$#e Possi$#e ?a#ues

    %. perate a Ban0 ime 5et2een customer 3 # arri!als in minutes(. est a ne2 chemical emperature < min %,##C %,# 3(%(

    process an ma3 (%(#C=

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    27/60

    5.4 Discrete Pro$a$i#ity Distri$utions

    he Pro$a$i#ity Distri$utionfor a ranom !aria5le escri5es ho2 pro5a5ilities areistri5ute o!er the !alues of the ranom !aria5le .

    Cor a iscrete ranom !aria5le 3" the Pro5a5ility Distri5ution is efine 5y apro$a$i#ity functionenote 5y f

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    28/60

    E'pecte- ?a#ue or mean of a ranom !aria5le is a measure of the central locationfor the ranom !aria5le

    ?ariance

    We use the Faria5le ?ariance in chapter & to summari?e the !aria5ility in ata. No22e use ?ariance to summari?e the !aria5ility in the !alues of a ranom !aria5le

    E'pecte- ?a#ue or mean of a Discrete ranom !aria5le:

    < = < =E x xf x= =

    ?ariance of a Discrete ranom !aria5le:

    ( (< = < = < =Var x x f x = =

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    29/60

    Some e'amp#es

    Using E'ce# to compute the E'pecte- ?a#ue; ?ariance an- Stan-ar- De@iation

    Eunction

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    30/60

    < =

    < =

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    31/60

    4. he occurrence or non-occurrence in any inter!al is inepenent of theoccurrence or non-occurrence in any other inter!al

    Poisson Pro$a$i#ity >unction

    < =J

    xef x

    x

    =

    2here

    is e3pecte !alue or mean num5er of occurrence in inter!al

    Some e'amp#es an- using E'ce# to compute Poisson Pro$a$i#ities

    < =J

    xef x

    x

    = :P11Nrom a set of n e#ements +hich contains r e#ements +ith the property A ta(e a

    samp#e N e#ements at ran-om. )et F $e the e'act#y ' e#ements +ith property A

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    32/60

    among the samp#e. F is ca##e- to $e a Hyper ,eometric Distri$ution ran-om

    @aria$#e

    Hypergeometric Pro$a$i#ity >unction*

    < =x n x

    r N r

    n

    N

    r N r

    x n xC Cf x

    NC

    n

    = =

    for # 3 r=

    Some e'amp#es an- using E'ce# to compute Hypergeometric Pro$a$i#ities

    < =x n x

    r N r

    n

    N

    r N r

    x n xC Cf x

    NC

    n

    = =

    :H4PEMD1

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    33/60

    8.1 The Pro$a$i#ity Density >unction

    Cor e!ery )anom !aria5le I " a funcion f

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    34/60

    N8E: Cor The Uniform Pro$a$i#ity Density >unction on inter@a# Ga;$

    < =(

    a bE x

    += an-

    (< =

    < =%(

    b aVar x

    =

    8./ Norma# Pro$a$i#ity Distri$ution

    Norma# Pro$a$i#ity Density >unction N; ( 3*

    (

    (

    < =

    (%

    < =

    (

    x

    f x e

    =

    Stan-ar- Norma# Pro$a$i#ity Distri$ution N2;13

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    35/60

    Stan-ar- Norma# Density >unction*

    (

    (%

    < =(

    x

    f x e

    =

    N8E: Area as a Measure of Pro$a$i#ity

    Because 2e ha!e < = < =P X f x dx

    = so" < =P X is the Area uner the cur!eof the f

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    36/60

    1uppose I has the Normal Pro5a5ility Distri5ution N< &'",## " ,##(= . Cin P< I R;#"### = " P< I ;#"###= G K

    Solution: We ha!e P< I R ;#"### = G P

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    37/60

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    38/60

    he Point Estimation of is(

    %

    < =

    %

    n

    i

    i

    x x

    sn

    =

    =

    he Point Estimation of Pro$a$i#ity p is fre&uency xn

    6./ SAMP)IN, DIST"IBUTI%NS

    he ranom !aria5les I%" I(" " In are aran-om samp#eof si?e n ifa he In>s are inepenent ranom !aria5les

    5 E!ery IL has the same pro5a5ility istri5ution

    A statisticis any function of the o5ser!ations in a ranom sample

    he pro5a5ility istri5ution of a statistic is calle a samp#ing -istri$ution

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    39/60

    f I%" I(" " In are a ranom sample of si?e n ta0en from a population 2ith

    mean an finite !ariance ( then

    X

    n

    = as n " is the stanar Normal

    istri5ution

    6./ Samp#ing Distri$ution of p

    he sample proportion p is the point estimator of the population p.

    E

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    40/60

    f np S , an n

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    41/60

    inter!al. Ho2e!er" the conUence inter!al is constructe so that 2e ha!e highconUence that it oes contain the un0no2n population parameter. 8onUenceinter!als are 2iely use in engineering an the sciences.

    =.1 C%N>IDENCE INTE"?A) %N THE MEAN %> A N%"MA)DIST"IBUTI%N; ?A"IANCE KN%!N

    If x is the samp#e mean of a ran-om samp#e of sie nfrom a norma# popu#ation

    +ith (no+n @ariance 4

    ; a 12213L CI on is gi@en $y

    A( A(A Ax ! n x ! n +

    !here A(! is the upper 1224 percentage point of the stan-ar- norma#

    -istri$ution.

    E'amp#e

    =.4 C%N>IDENCE INTE"?A) %N THE MEAN %> A N%"MA)

    DIST"IBUTI%N; ?A"IANCE UNKN%!N

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    42/60

    If x is the samp#e mean of a ran-om samp#e of sie nfrom a norma# popu#ation

    +ith un(no+n @ariance; a 12213L CI on is gi@en $y

    A( A(A Ax t s n x t s n +!here A(t is the upper 122 4 percentage point of the t -istri$ution +ith n1

    -egrees of free-om .

    E'amp#es

    =./ Determining the samp#e Sie

    If x is use- as an estimate of ; +e can $e 12213L conJ-ent that the errorx +i## not e'cee- a speciJe- amountE +hen the samp#e sie is

    (A(< =!

    n

    E

    ;

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    43/60

    E'amp#es

    =/ C%N>IDENCE INTE"?A) %N THE ?A"IANCE AND STANDA"D

    DE?IATI%N %> A N%"MA) DIST"IBUTI%N

    7et I%" I("" In5e a ranom sample from a normal istri5ution 2ith un0no2n!ariance ( . f s(is the sample then a 12213L confi-ence inter@a# on ( is

    ( (

    (

    ( (

    A(" % % A(" %

    < %= < %=

    n n

    n s n s

    2here(

    A(" %n an(

    % A(" %n are the upper an the lo2er %## A ( percentage points

    of the chi-s6uare istri5ution 2ith

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    44/60

    =.7 A)A",ESAMP)EC%N>IDENCEINTE"?A)>%"A

    P%PU)ATI%N P"%P%"TI%N

    f p is the proportion of o5ser!ation in a ranom sample of si?e n that 5elong to aclass of interest " an appro'imate 12213L confi-ence inter@a# on theproportion p of the popu#ationthat 5e long to this class is

    A( A(

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    45/60

    (A (< = #.(,

    !n

    E

    =

    Some e'amp#es

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    46/60

    CHAPTE" ai#ing to reOect the nu## hypothesisH2 +hen it is fai#se is -efine-as a type II error

    = Ptype I error3 : PreOectH2 +henH2 is true3

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    50/60

    Sometimes the type I error pro$a$i#ity is ca##e- the significance

    #e@e#; or the error; or thesie of the test.

    The po+er of a statistica# test is the pro$a$i#ity of reOecting the nu## hypothesis H2+hen the a#ternati@e hypothesis is true.

    T+o Tai#e- Tests*

    H2* # = H1* #

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    51/60

    his est is calle a t2o-sie test" 5ecause it is important to etect ifferences from thehypothesi?e !alue of the mean # that lie on either sie of #. n such a test" the criticalregion is split into t2o parts" 2ith

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    52/60

    the pro5lem conte3t.

    AN%"MA)DIST"IBUTI%N; KN%!N

    1uppose that 2e 2ish to test the hypotheses H#: # H%: # =

    Test Statistics#

    #

    X

    n

    =

    f the null hypothesis"#: # = is true" EIG # " an it follo2s that the istri5ution of#

    is the stanar normal istri5ution enoteN

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    53/60

    istri5ution of#2oul 5e unusual if"#: # = is true therefore" it is an inication that

    "# is false. hus" 2e shoul reLect"# if the o5ser!e !alue of the test statistic Y# is

    either #R

    -A(

    or#S A(

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    54/60

  • 8/11/2019 Bai Giangbusiness Statistics Hai Phong Dang Soan (1)

    55/60

    We may also e!elop proceures for testing hypotheses on the mean 2herethe

    a#ternati@e hypothesis is onesi-e-.1uppose that 2e specify the hypotheses as

    H#: # = H%: #