Assistente AI
Trascrizione
00:00:00Money in welcome to the first lesson experiment so.
00:00:10You remember that this is
00:00:13the second lesson su you know what is the first part
00:00:21in you make it right to answer and the second part will be
00:00:29a practical example in which we apply di analysis of two e che so,
00:00:38le start lesson and
00:00:43whiteboard you to your eyes about a nova or there is.
00:00:54IT something is not something.
00:01:00Il. Vincolo.
00:01:20Del pareggio.
00:01:28Di bilancio. Io dico.
00:01:31La. Slide non.
00:01:42So let's go to the.
00:01:48Let's non nova eggs.
00:01:55And. you return to. Slide number.
00:02:15To you to this sentence you so typically when you.
00:02:24Here we are speaking about.
00:02:28In What you have to have the effect of
00:02:32one single factor on air force typically when you have
00:02:38one single factor Ok you have to understand
00:02:43if that factor experts e influence on
00:02:49the science and influence
00:02:53is mindful from the statistical point of view.
00:02:56so you are going to perform physical test to end to end if.
00:03:03In the treatment about that factor
00:03:07or very high level of a single factor.
00:03:12You have a variation availability which which is much
00:03:18larger than the ability to have from
00:03:21the reputation of the same experiment.
00:03:25You have for you to know
00:03:28how so typically one you perform presentation for the time of.
00:03:42How to find the effect relation between input output when you.
00:03:52Make discrimination Gli typically make
00:03:56different reflections of the same experiment.
00:04:01So for your one of the principles of DUO-E you
00:04:07also retail in a nova
00:04:10is making applications of the same experiment.
00:04:14Ok, if you want to replicate an experiment for you have to me
00:04:21in the same conditions exactly the same conditions okay for more.
00:04:29A innova you have so this is what you have to keep uniform
00:04:38the conditions you have the system for you may have
00:04:42e system which is on the last of us,
00:04:47but for example, you are running e reactor Ok Reaction defence
00:04:54on the temperature ok. Catalyst
00:04:59utilizza un software but you are not interested in catalyst,
00:05:05you are only interested in standing for example if
00:05:10e provider of the material or another provider on
00:05:15the same material is going to
00:05:18the rest for example the
00:05:21yield of the reaction so you are keeping one factor,
00:05:26which is the provider of the year.
00:05:31Of your material and you want to this factor with.
00:05:39The yield of the reaction keeping uniform
00:05:44the conditions in your system is that
00:05:50Gli TALAPRO-2 design design catalyst the same overall system
00:05:59so di uniformità which is required year to the state of conditions.
00:06:08Ok. Ok. Allora questions
00:06:16ad alcuni quesiti.
00:06:57Travisare.
00:06:59Il. Senso.
00:07:10Della vita. Sorry. I get the.
00:07:16Ok, so let's go back here so typically.
00:07:25Noi di Napoli.
00:07:28Genova Noi Ok.
00:07:31He. He. We need to
00:07:38check IT is over the results so.
00:07:47Was what we live apart from the moon?
00:07:54What is this is what is the art of the model for example year.
00:07:59We are for example adoption model, which is.
00:08:04E. Means modo or alternative and effect model.
00:08:13Here we are.
00:08:18Our reasons di on the treatment.
00:08:24Okay, so the mind of the treatment so.
00:08:29Say. From the overall mean ok which is given by each treatment,
00:08:39but this is going to provide mirror
00:08:44and this errors should ever some nice court
00:08:50does caratteristiche are some raised in this slide so good good
00:09:00residuo normal should be Italy he should should
00:09:08bias should provide a bias in
00:09:12our model so they should be centered heroes.
00:09:19For example if you take in Instagram of the rules of your move.
00:09:29You should if this is trend, ovvero se continuo.
00:09:38Errors should be related to your noise white noise.
00:09:45So one of the noise is that it is.
00:09:53IT is fully random.
00:09:56So you have typically to understand
00:09:59if these noise diesel results are Randy
00:10:06and normally distributed so for
00:10:10all and you can perform or the normalità that you can perform
00:10:16they can be useful to access information so typically you have to
00:10:21understand this normal behaviour
00:10:25through normal probability close if you want to result to.
00:10:32Interpretation Ok as an alternative you can perform
00:10:38some formal socials Sacerdos
00:10:44in the first part of course for example in test.
00:10:52On the numerosi of you have.
00:11:00You are useful to understand that.
00:11:06Are. Were presented by
00:11:14your model so if you have to learn a residuo.
00:11:21If you find one of your observations.
00:11:25There is to large for this is
00:11:29an outlier disease e non invece in with your.
00:11:39Sono typically to understand if you have.
00:11:46You can result on.
00:11:51Files or student files Ok a standard.
00:12:01Sonic understand if the.
00:12:08Call that. You get your standard dice residuo di.
00:12:30Is your residuo by the standard.
00:12:40Of the story. So this should stay in the interval.
00:12:50Minus free free if you find the one.
00:12:59Is. Larger than three or lower the free.
00:13:10This means that this is very large
00:13:14and you have to understand if the point.
00:13:20He is not if something strange
00:13:25happen if some of the conditions in run
00:13:32the the system are not certified for that point so if you
00:13:39find something happen for your you have to remove this.
00:13:48Point and you have to.
00:13:57This condition.
00:14:03Not only residuo should have good behaviour in terms of.
00:14:12This behavior so se mi dite in distribution and so on the road
00:14:21show e random pattern
00:14:26so they should if you display your walls against time.
00:14:36So. And this is
00:14:41time so the time order in which you performing experiment.
00:14:46Okay. They should be something like
00:14:50this so random distributed going up and down.
00:14:54Ok, if you while if you find that you
00:14:59residuo time have some sequence some drift,
00:15:07which are evidence and behaviour which is not
00:15:11random once again means
00:15:15that there is an extra factor in your experiment,
00:15:20you don't take it to accounting
00:15:23and which is it peculiare type of ability.
00:15:30To indicate Io stufa
00:15:35diesel new factor ok
00:15:39and you have to identifying and the control it's.
00:15:44Not only you have to understand if there are
00:15:47some patterns in time what you have if
00:15:51those patterns can be found against
00:15:55the predictive value of white or white surf the real world wide.
00:16:04If you can find some strange powers when you.
00:16:13Build the place of the residence against.
00:16:20Why well?
00:16:24You are you are?
00:16:28Against white War against the War.
00:16:36On. This year you should have some random pattern Ok,
00:16:47if you find some strange pattern year old you should.
00:16:54Understand so for example if you find the pattern which is like
00:16:59this the black points years are not random distributed.
00:17:04Ok dei follow up pattern which is this banana Okay in this case
00:17:10something strange happens so the amount
00:17:14of the residuo dependence on.
00:17:20While and.
00:17:24Should be completely random finally.
00:17:31Here you need to understand if
00:17:37your residuo are what is this mean.
00:17:44Il test positivo a should display
00:17:50the same ability against the levels of the factors.
00:18:02So. If you display. Your residuo.
00:18:15Against example your factor
00:18:20the factor could be in a reactor, which you have to.
00:18:28Know the FLAURA2 ok amount of
00:18:32catalyst to keep all of this constant,
00:18:36but you want to understand
00:18:38if the real material provider
00:18:43as is equivalent to the real material provider B,
00:18:48and how this is the yield of your reaction.
00:18:54Ok, so you have as a factor.
00:18:59The raw material provider.
00:19:05And you have each and to different treatment that sei.
00:19:12Different providers here you have different.
00:19:21Observations in which you run your reactor and your Evaluate your
00:19:30evaluate your the year of
00:19:35the reaction when it is the material provided by provider.
00:19:41Then you have also some.
00:19:50Summer summer of this time
00:19:55when you the material provided by provider B.
00:20:02This is. This time out of our ability in
00:20:12one level one treatment and
00:20:17is provided by the other level the other treatment.
00:20:24Should be almost the same.
00:20:28And this La vin test or
00:20:37bullet test test if the
00:20:42ability within one of the levels.
00:20:51Factor or one of
00:20:55the treatment year experiment
00:20:59is the same as the treatment ability of the other treatment.
00:21:09Variety of the results of speaking about us so.
00:21:19That the spirit of the research.
00:21:26Indirizzo trick or in different levels of human factor at
00:21:36the same point of view and if you can
00:21:43see that the latest and test ok is test,
00:21:49which computers variant is which
00:21:55compare differences between the.
00:22:02Way you have.
00:22:08Which is reputation on treatment and the average of the treatment.
00:22:17In Vine. Mi fai delle intese?
00:22:21Io. Considera different reputation of an experiment.
00:22:31Within what treatment and compared this with
00:22:36the average medium conditions of
00:22:43the treatment within the ballad test you
00:22:46perform a formal test based on the square statistics.
00:22:52To understand if all the variant.
00:22:59You have in one to one. Of the factor.
00:23:08Starting is the same as you have in the other levels.
00:23:19IT is clear. Allora questo.
00:24:08Forse a volte anche chi non è scrittore.
00:24:17Può. No. Più Air Force
00:24:25but what you do is Billy in modo visivo.
00:24:35You're going to build. This model for example or
00:24:45small model or the effect
00:24:50more and then on the road to find on difference between.
00:24:58The treatment. And the real value you fine what is
00:25:07us and there is not should be did it they are not only.
00:25:16Seguiti sempre da un dio omicida.
00:25:20Stick if the year old does nice he is that we have.
00:25:32Been significant hanno ricevuto nella loro.
00:25:45Vita. A costo di norma un plus che.
00:25:57So you see that William treatment.
00:26:06Your meeting treatment this you are eating experiments. In which.
00:26:17di ritirare 100.000€ of
00:26:23death credo che sia io sono addirittura indietro.
00:26:31Dice You are only if the applications you do.
00:26:40Tra l'amore di Dio hanno un factors.
00:26:46Per la salute simili a.
00:26:5470. Anni il test ha.
00:27:03Mostrato di qualcosa che riguarda.
00:27:13Non solo in diesel.
00:27:18Test ok.
00:27:20If. You are in via novel table sei
00:27:30You are serving if
00:27:35Al termine o all treatment are the same.
00:27:42Or static different if the variant DUO-E
00:27:50to the treatment DUO-E to
00:27:53the variation of the level one factor for sample.
00:27:57Ok is the same has not.
00:28:03So here we are serving if the variabili given by
00:28:12the change of the dell'Aereo One factor
00:28:16or different three twenty is.
00:28:22The noise or it can be different much larger Okay so
00:28:29it is much larger year you have the that the observer is the.
00:28:37Way after that you are performing
00:28:44other types of tests and that are related to the facts that.
00:28:53Noise is normal distributed so this part of the model.
00:29:01The party describe.
00:29:05Di essi non.
00:29:10As good is only noise is
00:29:14only random noise because
00:29:18you don't want to that in this part of the past.
00:29:22In this.
00:29:26Sampling extra for you want that this is not so now we are.
00:29:35Expected in speaking.
00:29:39Diesel. So we want this
00:29:44is normally di we want this is.
00:29:53A talk show we have a point.
00:30:00We want to know if it shows
00:30:05some patterns in time or against the value
00:30:08or and we want to know if the error variabilità.
00:30:19Dipende. On the levels of the.
00:30:27Year we are understanding the ability within
00:30:33the world wide what we are doing year in dianova.
00:30:41We are serving if the variabili of
00:30:45the treatment is significant
00:30:49much larger than the world of the world.
00:30:52We are not watching simply within this.
00:30:57Okay, so fare sport di nuovo modo beauty.
00:31:04And the second is now I want to to see if also the noise as
00:31:14the caratteristici aspect and
00:31:18in this case too big to understand also if
00:31:22the ability of the noise within
00:31:27one level of the factors is to the one.
00:31:33We have others level of the other factors.
00:31:36This is not the same of the ability between levels of the factors.
00:31:44Ok. Giusto?
00:31:55Allora questions.
00:32:49Se sei.
00:33:00Un professore Example.
00:33:04We are second part of the lesson will be in example,
00:33:09which we will say sì
00:33:14all the steps of the procedure for building and testing.
00:33:20We are going to see in
00:33:23a moment so if you don't have any more questions,
00:33:27we can pass to the second life and so.
00:33:41WISE as starting point di nova
00:33:45so the objective year is to understand if different treatment
00:33:51of a single factor or different levels of
00:33:55a single factor to some differences on a response.
00:34:02To do Together with typically utilities e modo in with the rest of
00:34:12a different levels or
00:34:16different treatment and different petition
00:34:21of these different levels of.
00:34:27Different levels can you presented as the overall mean of
00:34:34the war place the time,
00:34:42which is what differenziate the mino
00:34:47from the middle class a San Salvador.
00:34:54Alcuni testi inediti.
00:35:00Di recente The way
00:35:05from other where we can do it in due alternative.
00:35:13One is an ipotesi Il test which
00:35:18is all the middle of all the treatment are the same.
00:35:23Ipotesi test against ipotesi at least for
00:35:30one treatment you have that
00:35:33the mean of the treatment is different from the other mean.
00:35:43This is particular valide for the mean model in with you say
00:35:51that way to new class there's ok,
00:35:59but if you consider is effect more Okay,
00:36:03you can translate this ipotesi testing in
00:36:07alternative ipotesi testing in which you have that Nell'ipotesi is
00:36:13that all the way of this time are not to zero
00:36:20so you don't have
00:36:22any treatment that differenziate way from the overall mean.
00:36:27Ok, ipotesi at least one of the time.
00:36:35Esiste invece that is not for
00:36:41you and you have been that treatment di francesi.
00:36:50Di miles of the right from the only.
00:36:59Dove ci si può presentare? Io vado.
00:37:04Allo stand e The total white you have your sister,
00:37:13Il main più noise.
00:37:19Overall. I dettagli decorativi.
00:37:26Il made in Italy DUO-E to the different nation of the treatment as
00:37:35the same making of the not so is not distinguished from north
00:37:42of the city to the facts that we more one factor on demand methods,
00:37:52we are the different treatment.
00:37:56Island much larger than
00:38:00the world wide web di nude simple reputation of.
00:38:08Displacement so once again,
00:38:13let's go to the example in which we have the way.
00:38:21And we have one factor here we have to treatment the.
00:38:30End of the same factor.
00:38:35We make some of the experiment that level
00:38:41eight summer of the experiment at level B.
00:38:49For sure we're going to observer it.
00:38:53Variabilità. Amore Petition.
00:39:04Of the same experiment this is not.
00:39:18You have. To take that you repeat the same,
00:39:26but you to some noise,
00:39:29you have your closed summer
00:39:31of the management and so on the sound of.
00:39:35This is love ability
00:39:39which is only go to the reputation of the experiment,
00:39:43which is only go to the noise of the process.
00:39:45You have while you have amount of war ability on why?
00:39:57Ucciso treatment which is to the take that you move.
00:40:03Your ADCs effect from the level hey to the level
00:40:09B. Siamo ancora aperti.
00:40:19I match larger than the amount of you have you to your error you to
00:40:28the reputation of the experiment under the same condition you
00:40:35to noise this means that this factor.
00:40:41This is the termine e large ability in why?
00:40:50E significant ability in white e variabili
00:40:56of which is significant larger than the bit.
00:41:01You have you to your error,
00:41:05but it may happen that you have that wearing your factor
00:41:16from eye to be the different treatment making
00:41:21some reputation for making some reputation
00:41:27for be okay you can see that the amount
00:41:34of ability you have in hay in treatment,
00:41:41you to your noise and out of you have.
00:41:46For the treatment be.
00:41:50Almost old ability to having your system.
00:42:01To your treatment year is providing to the way.
00:42:09Which is the same as you have.
00:42:21So of the treatment cannot be distinguished from the ability.
00:42:29To noise and in this case Okay,
00:42:34you are making test which
00:42:37comfort various and what is the test that is
00:42:43for comparing it is
00:42:48an So you have a new observer,
00:42:54which is the race between
00:42:58the balance you to the treatment to the variant due to the.
00:43:07End of this observer is within the limits.
00:43:15Of distribution with the number of the rise of freedom.
00:43:23And the number of freedom dominator.
00:43:28If this is true,
00:43:30this means that you cannot reject no hypothesis.
00:43:37While if you go in the region.
00:43:44You have to reject L'ipotesi torna l'ipotesi
00:43:49di that at least human Rights Watch
00:43:54is different from zero or at least one
00:43:59exists for which new age is different from.
00:44:05The other new.
00:44:19Year. You need is the rage between and
00:44:26the same square of the treatment
00:44:28by its number of decrease of freedom,
00:44:32which is the number of one.
00:44:36By the same square or by its number of freedom,
00:44:46which is the total number of experiments
00:44:48by the number of the treatment.
00:44:51Ok, so the war of the treatment of Dr.
00:44:57Dieci Larger than di confidence limits.
00:45:07Of confidence for the one of freedom on the dominator and
00:45:17the Handmaid's Tale denominatore degrees
00:45:22of freedom where is that you
00:45:24are in the reception on this is that
00:45:27the ability to the treatment is much larger than the variabilità.
00:45:33That the factor you very.
00:45:39The two level. In più si ripete uguale.
00:45:46Error. Of the noise
00:45:50so it is a influence e influence which is not simply noise.
00:46:01So what we have down then let's go to the.
00:46:08Moment of e ok,
00:46:14so di nuovo model in through.
00:46:17The computational effects.
00:46:22Performing the arts on the mind of the treatment on the way of
00:46:31the treatment with respect to the balance of the error
00:46:35you may find the year the list of the commons.
00:46:40You can we will see in
00:46:43practice not only you want to understand if if the treatment.
00:46:53If one factor experts e. Influence on this way,
00:47:02which is significant from the statistical point of view.
00:47:06What we can also what is this treatment,
00:47:10which is significant different from
00:47:13the other treatment in this case.
00:47:16You can it is the multiple comparison test success Typekit test or
00:47:22the method treatment which significant from the other treatment.
00:47:33From the other levels.
00:47:38And founder more.
00:47:42C'è modo e modo.
00:47:47You have to understand if there is
00:47:51what you have some pattern or not.
00:47:55If they are normal distributed if they have Anomalo.
00:48:00Anomalo. Values if if you
00:48:07can find some outliers in your your dataset.
00:48:13If they are affected by correlation if
00:48:17the same pattern against the original value.
00:48:22You can also perform a test intorno al test or
00:48:29living test ok reported
00:48:33the command MATLAB command for performance test.
00:48:38And then is possible.
00:48:41Lo studio empirico modelling because
00:48:44once again you have to be aware of the fact
00:48:48that ANOVA is the only opportunity to performance equality test,
00:48:56which is not thank you for influence on factor on the results.
00:49:05Ci sono segni di contractor,
00:49:10resources e significati. In questi.
00:49:22Giorni non sono molto noto.
00:49:25Molti dicono It's not want to find what
00:49:29is this given by the voice of the factor, ecc.
00:49:38So here we are building an empirico model.
00:49:47So. While anova model is to you rinvia.
00:50:00Il fatto che si verifichi un influencer.
00:50:07The results but if you want to have
00:50:12a quantitative description on air one factor the city of the.
00:50:23Two models we have to resort to multi linear model at based on
00:50:32your strategy minimale di immaginazione più che non essendo
00:50:41from the point of view
00:50:43what is it in the factors and responses
00:50:49through di simulation of simple regression Confessions.
00:50:58So while
00:51:01di nuovo modulo tela influence factor
00:51:06influence horizons di empirico
00:51:11modeling surface modeling profumati linear regression
00:51:16I was the opportunity to have.
00:51:19E così mi ha detto invece il matematico.
00:51:25Quantitativo inglese which is
00:51:29what is the influence of the factor on the.
00:51:35Reasons why best from it to what if from
00:51:42if one factor influence on response
00:51:47to what if the influence of the factor to the.
00:51:52Place to the first example is it to
00:51:58the manufacturing of integrated circuit for example year
00:52:02we are manufacturing so what is
00:52:06this is which is a collection of integrated circuit.
00:52:12Ok. IT is really complex to to make this way for
00:52:19what is typically done is
00:52:24performing several times type of procedure,
00:52:29so you have typically e semiconductor.
00:52:32Ok, semiconductor in you put some.
00:52:40Layer of coating scouting is sensitive to
00:52:47light so what is that you like you have a mask.
00:52:56Like this Ok. E mask in which you should some light.
00:53:03Okay, so the light doesn't pass years pass
00:53:07year and years ok and so on the light roads.
00:53:12This in this point giving e peculiare shape which is this.
00:53:25Is so your life pass roads dice Ok,
00:53:34and you have this shake then
00:53:40after light exposure and development Ok you are some yes year.
00:53:50Some plasma and his.
00:53:56Make this semiconductor.
00:54:00He developed different sounds.
00:54:06Simon, which are highly conductor.
00:54:10This is not conduct so you clean everything ok and you have
00:54:19this semiconductor in which you have
00:54:23some of the use of Non connectivity.
00:54:31Ok, so.
00:54:35Put your weather ok within the elections
00:54:42del mercato ok and used summer radio frequency generator to.
00:54:50Transform this gas into a plasma which
00:54:55modified the surface characteristic of this semiconductor
00:55:01is what you want to understand is the relation between
00:55:09the power of the radio frequency to
00:55:12convert discussion a plasma which
00:55:15can penetrate in semiconductor years.
00:55:18And. So you have to perform
00:55:28file options over at for different levels of
00:55:38the power for the frequency so and power of the world
00:55:48and you perform this experiments five times in in
00:55:55Italy indipendent member in a random order
00:55:59with the other three levels.
00:56:03You have you need for the power of the video frequency.
00:56:09Sono. Nati suonando dentro.
00:56:17Casa total year for.
00:56:22Free events for levels of your factor and the Five.
00:56:31Euro. So userform total of
00:56:37twenty experiments five replicate
00:56:43for each of the four leavers in you move your factor,
00:56:49the power of the frequency to understand which is this number.
00:56:59So you can find z in the good wife and scuola nuova.
00:57:09Now we are going to build as
00:57:15usual a preliminary visualizza agent of the.
00:57:21Effect per.
00:57:24Videoscrittura su power export on, ecc.
00:57:29Durante il mese in.
00:57:33Inglese. So what is
00:57:40that the power of the expert?
00:57:51Significa. In sold out cosicché.
00:58:03Così lo so solo io.
00:58:08Ha un blog in più il download on
00:58:14the job for the last of this network.
00:58:22Ok, gli ultimi dati di.
00:58:29Fatto. We don't have any secondo significant we want to know if
00:58:37our feeling about this is that the power reserve is on the road.
00:58:46Ci sono un sacco di pay.
00:58:50To you want to perform a una nuova.
00:58:56Spesa. We want to know
00:59:00if they are also meaningful from the statistical point of
00:59:03view so we are going to build in a nuova model to build.
00:59:09Un nuovo model. You can for one single factor Ok go for a one way
00:59:17anova a mo Which understand di effect of one single factor on.
00:59:28Through the command in one.
00:59:36You can see that this command give output bob ti.
00:59:47Posso dire di sì e. A volte.
00:59:55Sì. What is the effect of the different Columbus Day?
01:00:03Effect of this is what we can
01:00:11see the effect of different treatment
01:00:15different power of the radio frequency.
01:00:19ok and in the year you can see the white house,
01:00:26which is the transport of these methods.
01:00:31We have on the claims
01:00:33different treatment different levels in the white house,
01:00:39okay and year you can see the samos of
01:00:44the same square of the land
01:00:47the treatment compared with the same square of.
01:00:54The total same where.
01:00:58You can see the number of the rise of freedom for
01:01:02each of us so you have anything degrees of freedom.
01:01:12Which are the free Di solito la society
01:01:21for treatment minus one in one and many sei denaro.
01:01:32Minus for present sixteen degrees of freedom,
01:01:38which associated to the error so if you.
01:01:43Divided square by the number of degrees of freedom of time square.
01:01:51How to treatment which is thousand and the old error,
01:01:58which is not for meaning that
01:02:03there which contribute that is this point.
01:02:11And the point said for well within the reception sono.
01:02:20Using p value which is to point something then to the times,
01:02:27then to the power of minus nine so it is really
01:02:31to match lower point five, which is speaking.
01:02:37Providing as a number
01:02:41that it is arrived from
01:02:44the statistical point of view that the war of
01:02:47the treatment is much larger than the war of the ability to
01:02:53the accurate by the variation of the radiofrequenze power.
01:03:03Il match larger than you have you to the simple reputation of.
01:03:13So you that the ability from one to one twenty Ok,
01:03:23it's from five hundred.
01:03:26Certi forti two seven,
01:03:30twenty way meeting, the same experiment,
01:03:35you have of he non so,
01:03:41I have you have which is he.
01:03:45Ci vuole un applauso.
01:03:51You have in the domain or from five to save
01:03:57the way in treatment
01:04:02is from for under the night to five hundred.
01:04:11Five. Servito Fifty compared to to
01:04:17hundred to the city
01:04:22of the reputation of one experiment is much lower than the city,
01:04:27you have in entire range of.
01:04:33This is that disability in the ability to have the way
01:04:40di ADCs effect Il DUO-E to the variation of
01:04:45the frequency power is much larger than
01:04:48the city web reputation of experiments.
01:04:52Web. So if you find that this is more than zero point zero five
01:05:03typically Gli consider that this is mindful Vincoli influenzino.
01:05:14Temporalmente o vuoi.
01:05:23Gli.
01:05:33Ok, so with this command one Ok you can put it.
01:05:41In ogni input is this.
01:05:45Matrix with the different factors in our providing both wise
01:05:55so providing e vector which is
01:06:01that for example to you have the five,
01:06:08but also for each and you your.
01:06:13Five hundred and so on and support and you can
01:06:18put in this alternative form and you can do it.
01:06:26If you want by yourself so we the rise of code
01:06:30you can computer the total square on your way,
01:06:37you can commit treatment.
01:06:41You can error which is
01:06:45the difference among the previous you can commit the.
01:06:52Number the number of the rise of freedom
01:06:55to treatment and total what.
01:06:59You can comp the times of the year you can d f. Which is
01:07:09the real between the mean square
01:07:14of the year of the year and compared with the.
01:07:21Limit for an effect.
01:07:28Which is limit to.
01:07:32Lose beyond this limit or you can can be thoughts on
01:07:38the value from the year of distribution with the house of.
01:07:49All you can find all the numbers you have from the year,
01:07:57you obtaining the value and the table of the statistics.
01:08:09Not only you have to understand one factor,
01:08:16the power of
01:08:18power resort e mindful influence online influenzano il.
01:08:27Modo con sottigliezza del he
01:08:33different treatment here we have four for levels.
01:08:41From the other so what you can perform a tale testo ad altri testi.
01:08:50Ora, in Test you want to pay TV the good Malcom
01:08:57per you make compreso off on the road.
01:09:02Quindi alla fine, e mi dispiace che io ho tentato di.
01:09:10Violare il fine.
01:09:15The ability within the first created.
01:09:18We are the one of which is the power of the world.
01:09:26Okay, ti lamenti di Cicerone and fifty over dieci?
01:09:34Okay. Which is where distinguished from the other.
01:09:40Responses. To this is on the law.
01:09:51Of a significant different from
01:09:56the others sono that the Meglio così.
01:10:04Power expert e influence on the wind
01:10:12which you with a dirmi cosa?
01:10:22Devo fare? What you know better and the tribes?
01:10:29All the best of the factor.
01:10:33Il termine availability in.
01:10:38From the mind.
01:10:42Which is your levels.
01:10:48Ovvero. So we know that there is
01:10:57better frequency power Esorta l'influencer Meaningful on the
01:11:06what we know that all the level
01:11:10of that factor are fully different from.
01:11:17The effect of the other levels through
01:11:20the this test now we have to check the residence so we can
01:11:27commit dice from the difference between the real value and
01:11:36the mind of the treatment
01:11:39and we can build normal probability plot of the year.
01:11:45You can see that diesel residuo are Chandler down.
01:11:50And all there is what are close to
01:11:56the right line from this is So.
01:12:04There are normally distributed according to
01:12:09probability plot if you want you can perform
01:12:13a test and the evidence that we need is that the.
01:12:26Our reviews.
01:12:29Non è difficile il compito The standard as.
01:12:38We can see that they say between
01:12:41minus one point five and one point five.
01:12:45So what is the three free.
01:12:49Limits mean that is now outlier in our budget or the data
01:12:56good ok Riprese di tipo model and
01:13:02if the residuo di observations there is it was against the way.
01:13:12Of the treatment while that the.
01:13:19Good version of that points.
01:13:25Information and we can test if the rise of city through
01:13:33the test or the test which gives epi value,
01:13:41which is much larger than zero point of five mind that
01:13:45the rise of the ability within the.
01:13:52Level of the treatment is
01:13:55the same La bici viene inviata
01:13:58a You can see you from year to year of
01:14:03the city of the world within
01:14:07one treatment is the same other levels.
01:14:19Year you have the.
01:14:23Evidence from the statistical point of view.
01:14:29If you go to Matlab can be very easily so let's go.
01:14:48Let's go to the dataset it provided.
01:14:54Dis dataset.
01:14:56Ok, so what a nova here we can input the data in Matlab.
01:15:04Ok, as you can see you
01:15:08have for for different treatment on the road.
01:15:17We have five reputation.
01:15:21This is the first row for the treatment one hundred sixty what one.
01:15:29Two hundred e hundred.
01:15:37Training you can be what you want.
01:15:41Box plot of white transport.
01:15:50Transport.
01:15:54And. You know what is now?
01:16:02Okay, if you can see that at the number one, which is one hundred.
01:16:08Walt Whitman number two which is what if you have for different.
01:16:17Plots which related your frequency radio frequency power level
01:16:28to the way to the rate on the year.
01:16:34We can that the significant influence of the factor on the rest,
01:16:41but we want e. Of this significant so we can build e in one.
01:16:52Modo. So with this command page table stats.
01:17:01Ok. Di nuovo one modulo on UI transport if
01:17:06we build 19 module here we are we have both the box ready.
01:17:14To reset and.
01:17:16Di nuovo table with the treatment on on the columns.
01:17:23Ok. IT is that we have a petition for treatment on
01:17:29fire petition for treatment and so on and so forces
01:17:32that is the treatment one is this on.
01:17:41The mine between five.
01:17:50Poi vanno inseriti il five fifty one point
01:17:55you to this is the way presented.
01:18:01The first and the one of the world of
01:18:06power and these are the variations.
01:18:14With respect to this.
01:18:19Average. Di average in the second treatment what is
01:18:28an eight seven and also the average
01:18:33of the treatment at hundred what.
01:18:39Is found years in the stars,
01:18:43so if you go to stars and to means year you can
01:18:49find the way of the model you know that we have a modo.
01:18:56E modo with the place that we can say
01:19:02that it is the new age plus the.
01:19:11Year you have the new age.
01:19:14Of the mind of the.
01:19:21Twenty. What's while for each of this?
01:19:29We have the ability so for example you can show this in
01:19:34a form to this is this is my way,
01:19:40sorry so for the first experiment and we have a five.
01:19:47Seventy five as the second for the second.
01:19:51We have five hundred and for you for the third petition.
01:19:57We have. The petition and the fifth Petition. We have.
01:20:08The end.
01:20:14Of the year.
01:20:17We have the variation
01:20:19around the five hundred fifty one which is the man of death.
01:20:30Year in full automatic matter.
01:20:34We have the mutation of the same square of
01:20:40the treatment square of
01:20:43the error the number the Number of freedom and the war,
01:20:49which is the same divided by the respect number of freedom.
01:20:54So we can commit di observer that which is
01:20:58the race between the two square and we can commit this value.
01:21:03If we want we can do it in a for
01:21:07independent manner ok year and also year in the.
01:21:16Slides reported The fool.
01:21:23Code Okay, so if you want to we can run this ok with one command
01:21:34in the two versions with the white matrix only or with why Ok Nel
01:21:44forum If you want to we can do it coming out
01:21:50of the total square square treatment
01:21:54and square of the year of freedom,
01:21:57the computational of the square of
01:21:59the treatment square of the error computational of the house
01:22:03of the night percent confidence limit of distribution and value.
01:22:12Can do not so perform
01:22:16a multiple comparison tests so once we
01:22:19know what the value is not so you are
01:22:23you can find your mind P p value
01:22:28which is much lower than zero point five meaning that we
01:22:32have a meaningful difference between the variant
01:22:37given by the treatment and balance you to your noise.
01:22:42Ok here we can say that we have an evidence
01:22:47that the treatment resort and influence on the residence,
01:22:52but old levels of
01:22:56the treatment different or
01:22:57not can performance multiple comparison test.
01:23:10With this command and here we are we have this plot.
01:23:19Multiple comparison test less.
01:23:26Is this is ok and.
01:23:34We are if you go to
01:23:37this level which is the first treatment Buon Android.
01:23:40What if you go to this level which is
01:23:44the second treatment at the level treatment out hundred
01:23:50and level you can see that the levels
01:23:54are well distinguished from the other levels no one
01:23:58overlap with the other meaning that all the levels of
01:24:04that factors given responses to give availability of the responses,
01:24:12which is what distinguished from the visibility given to
01:24:17the way by all the other levels of this of the same factor.
01:24:24Ok. Ratings.
01:24:31That we run this God.
01:24:34Ok, Let's run if you want if you run this code all in
01:24:41one year a implementi also
01:24:43all the computational of the residuo is the standard.
01:24:50And the figures not only from the graphic point of view,
01:24:57but also from the statistical point of view.
01:25:00We have the latest test to the.
01:25:05Test on the.
01:25:09On the of the city of the year.
01:25:18You can see that you have
01:25:22the multiple tests ok it in different manner.
01:25:30Ok, you have that this is the value associated della test
01:25:36match laser point five we have the value of the bullet test,
01:25:44which is your point free much larger than zero zero five.
01:25:48So we have an evidence city
01:25:51and we have all the graphic interpretation about this
01:25:55is so this is which is as well within the minus free levels,
01:26:03which is that we don't have any outliers.
01:26:07We have a normal probability plot,
01:26:11which is that they are normal distributed.
01:26:14We have the study of the results in time
01:26:19the war against the average of the treatment and year
01:26:24you can see a. Grafico dimostrativo di una serie di designs
01:26:29that year of the sound of the same at the levels.
01:26:35Ok. you have the best way and that's it.
01:26:47Was. Okay, so what is.
01:26:57This lesson ok a year of the year linear model is not
01:27:06so so good can be much better even if you may have.
01:27:16Speak about. Modeling and surface modeling in hours.
01:27:29Thank you and see you later.