Aop 1.Kol
.
- 1.
U kriterijumu sume kvadrata gresaka u klaster analizi, greska se odredjuje:
- A.
Kao rastojanje svakog vektora obelezja do centra najblizeg klastera
- B.
Kao rastojanje do vektora srednje vrednosti za sve raspolozive podatke
- C.
Kao rastojanje svakog vektora obelezja do centra najdaljeg klastera
- D.
Kao rastojanje do najblizeg vektora obelezja
Correct Answer
A. Kao rastojanje svakog vektora obelezja do centra najblizeg klasteraExplanation
In the criterion of sum of squared errors in cluster analysis, the error is determined as the distance of each feature vector to the center of the nearest cluster. This means that the error is calculated by measuring the distance between each data point and the centroid of the cluster it belongs to. The closer the data point is to the centroid, the smaller the error will be.Rate this question:
-
- 2.
Razdvajajuce hijerarhijsko klasterovanje pocinje tako sto su:
- A.
Svi vektori obelezja objedinjeni u jedan klaster
- B.
Svaki vektor obelezja je jedan klaster
- C.
Svi vektori obelezja podeljeni u unapred zadati broj klastera
Correct Answer
A. Svi vektori obelezja objedinjeni u jedan klasterExplanation
In divisive hierarchical clustering, all feature vectors are initially merged into one cluster. This means that at the beginning, there is only one cluster that contains all the feature vectors.Rate this question:
-
- 3.
Sekvencijalni podaci se odlikuju po tome sto je:
- A.
Redosled nije bitan
- B.
Ne menjaju znacenje ako se permutuju
- C.
Redosled bitan
Correct Answer
C. Redosled bitanExplanation
Sequential data is characterized by the fact that the order or sequence of the data is important. This means that changing the order of the data can alter its meaning or interpretation. Therefore, in this case, the correct answer is "Redosled bitan" which translates to "Order is important."Rate this question:
-
- 4.
Agregirani podaci imaju tendenciju:
- A.
Imaju veca odstupanja
- B.
Ostanu isti u pogledu odstupanja u odnosu na neagregirane podatke
- C.
Imaju manja odstupanja
Correct Answer
C. Imaju manja odstupanjaExplanation
Agregirani podaci imaju tendenciju da imaju manja odstupanja. This means that when data is aggregated, the variations or differences between the individual data points are reduced. Aggregation involves combining multiple data points into a single value or summary, which can help to smooth out any outliers or extreme values. Therefore, the aggregated data tends to have smaller deviations or discrepancies compared to the unaggregated data.Rate this question:
-
- 5.
Objekat je u nauci o podacima isto sto i:
- A.
Instanca
- B.
Primer
- C.
Slucaj
- D.
Atribut
- E.
Tacka
- F.
Obelezje
- G.
Entitet
- H.
Broj
- I.
Osobina
- J.
Slog
Correct Answer(s)
A. Instanca
B. Primer
C. Slucaj
E. Tacka
G. Entitet
J. SlogExplanation
The correct answer is "Instanca, Primer, Slucaj, Tacka, Entitet, Slog." In the field of data science, an object refers to a specific occurrence or example of a concept or entity. It can be represented by various terms such as instance, case, point, entity, or record. These terms are used interchangeably to describe a unique unit of data that is being analyzed or processed.Rate this question:
-
- 6.
Na intervalne atribute se moze primeniti operacija:
- A.
Razlicitosti
- B.
Uredjenja
- C.
Aditivnosti
- D.
Multiplikativnosti
Correct Answer(s)
A. Razlicitosti
B. Uredjenja
C. AditivnostiExplanation
The correct answer is "Razlicitosti, Uredjenja, Aditivnosti". This means that the operation of "Razlicitosti" (inequality), "Uredjenja" (ordering), and "Aditivnosti" (additivity) can be applied to interval attributes. These operations allow for comparing, ordering, and performing addition on interval values, respectively.Rate this question:
-
- 7.
Vizualizacija podataka je:
- A.
Veoma znacajna za nauku o podacima
- B.
Znacajna je samo za dvodimenzionalne podatke
- C.
Nije relevantna za nauku o podacima
Correct Answer
A. Veoma znacajna za nauku o podacimaExplanation
Visualization of data is very important for data science because it allows researchers and analysts to easily understand and interpret complex patterns and relationships within the data. By representing data visually, it becomes easier to identify trends, outliers, and anomalies, which can lead to valuable insights and informed decision-making. Additionally, visualizations can help communicate findings to stakeholders in a clear and intuitive manner, enhancing the overall impact and effectiveness of data analysis.Rate this question:
-
- 8.
Na redne atribute se moze primeniti operacija:
- A.
Multiplikativnosti
- B.
Aditivnosti
- C.
Uredjenja
- D.
Razlicitosti
Correct Answer(s)
C. Uredjenja
D. RazlicitostiExplanation
The correct answer is "Uredjenja, Razlicitosti." This means that the operation of ordering and the operation of distinctness can be applied to ordinal attributes. In other words, ordinal attributes can be arranged in a specific order and can have different values.Rate this question:
-
- 9.
Tipovi atributa su:
- A.
Informativni
- B.
Prostorni
- C.
Redni
- D.
Numericki
- E.
Razmerni
- F.
Intervalni
- G.
Imenski
- H.
Sekvencijalni
Correct Answer(s)
C. Redni
E. Razmerni
F. Intervalni
G. ImenskiExplanation
The given options represent different types of attribute scales. "Redni" refers to ordinal scales, which represent a ranking or order of values. "Razmerni" refers to ratio scales, which have a fixed zero point and allow for meaningful ratios between values. "Intervalni" refers to interval scales, which have equal intervals between values but no fixed zero point. "Imenski" refers to nominal scales, which represent categories or labels without any inherent order or numerical value.Rate this question:
-
- 10.
U modelovanju klasifikacionog problema, smisao ekskluzivnosti klasa znaci:
- A.
Klase se preklapaju
- B.
Klase su multiplikativne
- C.
Klase se ne preklapaju
- D.
Klase su aditivne
Correct Answer
C. Klase se ne preklapajuExplanation
The correct answer is "Klase se ne preklapaju" which translates to "Classes do not overlap" in English. This means that the classes in the classification problem are distinct and do not have any common elements. Each data point belongs to only one class and there is no ambiguity or overlap in assigning the data points to their respective classes.Rate this question:
-
- 11.
Mahalanobisovo rastojanje je korisno kada vazi:
- A.
Atributi nisu korelisani
- B.
Raspodela atributa je proizvoljna
- C.
Atributi su kontinualni
- D.
Atributi imaju iste opsege
- E.
Atributi imaju razlicite opsege
- F.
Raspodela atributa je priblizno Gausova(Normalna)
- G.
Atributi su korelisani
Correct Answer(s)
E. Atributi imaju razlicite opsege
F. Raspodela atributa je priblizno Gausova(Normalna)
G. Atributi su korelisaniExplanation
Mahalanobis distance is useful when the attributes have different ranges, the distribution of attributes is approximately Gaussian (Normal), and the attributes are correlated.Rate this question:
-
- 12.
Sta je verodostojnost (likelihood) klase omega_i:
- A.
P(X|omega_i)
- B.
P(omega_i|X)
- C.
P(omega_i)
Correct Answer
A. P(X|omega_i)Explanation
The given answer, P(X|omega_i), refers to the likelihood of class omega_i given the data X. This term represents the probability of observing the data X, given that the true class is omega_i. In other words, it quantifies how well the data X is explained by the class omega_i. It is an essential component in various statistical and machine learning algorithms, such as Naive Bayes and Maximum Likelihood Estimation, for estimating class probabilities and making predictions.Rate this question:
-
- 13.
Prokletstvo dimenzija je pojava pri kojoj se:
- A.
Pri povecanju dimenzionalnosti podataka nuzno javlja njihovo proredjivanje u prostoru koji zauzimaju
- B.
Podaci udaljuju od koordinatnog pocetka
- C.
Podaci pomeraju ka koordinatnom pocetku
- D.
Pri povecanju dimenzionalnosti podataka nuzno javlja njihovo zgusnjavanje u prostoru koji zauzimaju
Correct Answer
A. Pri povecanju dimenzionalnosti podataka nuzno javlja njihovo proredjivanje u prostoru koji zauzimajuExplanation
When the dimensionality of data increases, it is necessary for the data to become more sparse in the space they occupy. This means that the data points become more spread out and move away from the coordinate origin. As the dimensionality increases, the data points tend to become more scattered and spread out, leading to sparsity in the space they occupy.Rate this question:
-
- 14.
U klasifikacionim problemima, smisao potpunosti klasa znaci:
- A.
Klase cine jedan partitivni skup u prostoru klasa
- B.
Klase su aditivne
- C.
Svaki uzorak koji podleze klasifikaciji mora pripadati jednoj od definisanih klasa
- D.
Klase su medjusobno nezavisne
Correct Answer
C. Svaki uzorak koji podleze klasifikaciji mora pripadati jednoj od definisanih klasaExplanation
In classification problems, the concept of class completeness means that every sample that undergoes classification must belong to one of the defined classes. This implies that there should be no samples that do not fall into any class or belong to multiple classes. Class completeness ensures that all samples are properly categorized and accounted for in the classification process.Rate this question:
-
- 15.
Agregacija je operacija:
- A.
Eliminacije objekata
- B.
Kombinovanja dva ili vise atributa
- C.
Kombinovanja dva ili vise objekata
- D.
Eliminacije atributa
Correct Answer(s)
B. Kombinovanja dva ili vise atributa
C. Kombinovanja dva ili vise objekataExplanation
Aggregation is an operation that combines two or more attributes or objects. It is a process of creating a new entity by combining existing attributes or objects. In this case, the correct answer includes both options of combining two or more attributes and combining two or more objects.Rate this question:
-
- 16.
Atributi su u nauci o podacima isto sto i:
- A.
Objekti
- B.
Karakteristike
- C.
Tacke
- D.
Vektori
- E.
Svojstva
- F.
Slogovi
- G.
Obelezja
- H.
Instance
- I.
Primeri
Correct Answer(s)
B. Karakteristike
E. Svojstva
G. ObelezjaExplanation
The correct answer is "Karakteristike, Svojstva, Obelezja." In the field of data science, "atributi" refers to the characteristics or properties of objects or instances. These attributes can be used to describe or differentiate different data points or objects. Therefore, "Karakteristike, Svojstva, Obelezja" are all synonyms for "atributi" in this context.Rate this question:
-
- 17.
U svakom koraku rada razdvajajuceg hijerarhijskog klasterovanja:
- A.
Klasteri se dele
- B.
Prvo se podele, a zatim se objedine
- C.
Klasteri se objedinjavaju
- D.
Prvo se objedine, a zatim se dele
Correct Answer
A. Klasteri se deleExplanation
In each step of divisive hierarchical clustering, clusters are divided. This means that the existing clusters are split into smaller clusters based on some criterion, such as distance or similarity. This process continues until each data point is assigned to its own individual cluster.Rate this question:
-
- 18.
Da li je u hijerarhijskom klasterovanju neophodno unapred odrediti broj klastera?
- A.
Da
- B.
Ne
Correct Answer
B. NeExplanation
In hierarchical clustering, it is not necessary to predefine the number of clusters. This is because hierarchical clustering builds a tree-like structure (dendrogram) where the number of clusters can be determined by cutting the dendrogram at a desired height. This allows for flexibility in choosing the number of clusters based on the specific requirements of the analysis. Therefore, the correct answer is "No".Rate this question:
-
- 19.
Na koji nacin se nakon dobijanja dendograma u hijerarhijskom klasterovanju dobija zeljeni broj klastera?
- A.
Skracivanjem dendograma na odgovarajuci nivo
- B.
Reklasifikacijom
- C.
Eliminacijom uzoraka
Correct Answer
A. Skracivanjem dendograma na odgovarajuci nivoExplanation
Nakon dobijanja dendograma u hijerarhijskom klasterovanju, zeljeni broj klastera se dobija skracivanjem dendograma na odgovarajući nivo. Ovo se može postići postavljanjem praga na odgovarajuću visinu u dendogramu, gde se presecaju grane koje predstavljaju klasterovanje. Na taj način se formiraju klasteri prema zadanom broju.Rate this question:
-
- 20.
Tipovi uzoraka su:
- A.
Izbor uzorka po slojevima (delovima)
- B.
Semideterminisani
- C.
Izbor uzorka sa zamenom
- D.
Preferencijalni
- E.
Jednostavni slucajni uzorak
- F.
Izbor uzorka bez zamene
- G.
Didakticki
Correct Answer(s)
A. Izbor uzorka po slojevima (delovima)
C. Izbor uzorka sa zamenom
E. Jednostavni slucajni uzorak
F. Izbor uzorka bez zameneExplanation
The correct answer includes different types of sampling techniques. "Izbor uzorka po slojevima (delovima)" refers to stratified sampling, which involves dividing the population into different strata or layers and then selecting a sample from each stratum. "Izbor uzorka sa zamenom" refers to random sampling with replacement, where each member of the population has an equal chance of being selected multiple times. "Jednostavni slucajni uzorak" refers to simple random sampling, where each member of the population has an equal chance of being selected. "Izbor uzorka bez zamene" refers to random sampling without replacement, where each member of the population can be selected only once. These different sampling techniques allow researchers to obtain representative and unbiased samples for their studies.Rate this question:
-
- 21.
Egzogene metode redukcije dimenzija podataka:
- A.
Minimizuju diskriminatornu informaciju unutar podataka
- B.
Maksimizuju diskriminatornu informaciju unutar podataka
- C.
Maksimizuju informaciju o skupu podataka kao celini
- D.
Minimizuju informaciju o skupu podataka kao celini
Correct Answer
A. Minimizuju diskriminatornu informaciju unutar podatakaExplanation
The correct answer is "Minimizuju diskriminatornu informaciju unutar podataka". This means that exogenous dimension reduction methods aim to minimize the discriminative information within the data. These methods are used to remove features or dimensions that do not contribute significantly to the classification or clustering of the data, thereby reducing the complexity and improving the efficiency of data analysis algorithms. By minimizing the discriminative information, these methods help in reducing redundancy and noise in the data, leading to better data representation and analysis.Rate this question:
-
- 22.
Klasterovanje pomocu K-sredina je:
- A.
Mesovito klasterovanje
- B.
Hijerarhijsko klasterovanje
- C.
Particiono klasterovanje
Correct Answer
C. Particiono klasterovanjeExplanation
The correct answer is "Particiono klasterovanje." This refers to partition clustering, which involves dividing the data into non-overlapping subsets or partitions. Each partition represents a cluster, and each data point belongs to only one cluster. This method is different from hierarchical clustering, where clusters are formed based on a hierarchy of nested clusters, and from mixed clustering, which combines different types of clustering algorithms.Rate this question:
-
- 23.
Da li izbor pocetnih centroida u algoritmu K-sredina utice na konacno resenje?
- A.
Nekada utice, nekada ne utice
- B.
Utice
- C.
Ne utice
Correct Answer
B. UticeExplanation
The correct answer is "Utice" which means "It does influence" in English. This implies that the choice of initial centroids in the K-means algorithm does have an impact on the final solution. The initial centroids determine the starting points for the clustering process, and different initial choices can lead to different cluster assignments and ultimately different final solutions.Rate this question:
-
- 24.
Klaster analiza je pronalazenje grupe objekata takvih da su:
- A.
Objekti u jednoj grupi slicni i da su istovremeno razliciti od objekata u drugim grupama
- B.
Objekti u razlicitim grupama razliciti
- C.
Objekti u jednoj grupi slicni
Correct Answer
A. Objekti u jednoj grupi slicni i da su istovremeno razliciti od objekata u drugim grupamaExplanation
The correct answer suggests that cluster analysis involves finding groups of objects that are similar within each group but different from objects in other groups. This means that objects within a cluster share similarities among themselves but are distinct from objects in other clusters.Rate this question:
-
- 25.
Kod aglomerativnog hijerarhijskog klasterovanja, na pocetku rada je:
- A.
Svaki vektor obelezje je jedan klaster
- B.
Svi vektori su jedan klaster
- C.
Svaki drugi vektor obelezja je jedan klaster
Correct Answer
A. Svaki vektor obelezje je jedan klasterExplanation
At the beginning of agglomerative hierarchical clustering, each vector feature is considered as a separate cluster. This means that initially, each vector is treated as its own cluster. As the clustering algorithm progresses, these individual clusters will be merged together based on their similarity until a final clustering solution is obtained.Rate this question:
-
- 26.
Jednostavan model sa puno podataka za obucavanje je bolji od slozenijeg modela sa manje obucavajucih podataka:
- A.
Netacno
- B.
Nije moguce takvo poredjenje
- C.
Tacno
Correct Answer
C. TacnoExplanation
A simple model with a large amount of training data is better than a complex model with fewer training data because having more data allows the model to learn more effectively and make more accurate predictions. The simplicity of the model also helps in reducing overfitting and improves generalization. Therefore, the statement "Tacno" (True) is correct.Rate this question:
-
- 27.
Sta je evidens?
- A.
P(X)
- B.
P(omega_i|X)
- C.
P(X|omega_i)
Correct Answer
A. P(X)Explanation
The correct answer is P(X). In probability theory, P(X) represents the probability of event X occurring. It is a measure of the likelihood of X happening. In this context, "evidens" refers to the evidence or information available. Therefore, P(X) represents the probability of event X given the available evidence.Rate this question:
-
- 28.
Na razmerne atribute se moze primeniti operacija:
- A.
Aditivnosti
- B.
Multiplikativnosti
- C.
Razlicitosti
- D.
Uredjenja
Correct Answer(s)
A. Aditivnosti
B. Multiplikativnosti
C. Razlicitosti
D. Uredjenja -
- 29.
Ako je diskriminaciona funkcija klase i data sa g_i_(x), i=1,2...,C, tada vazi: ???
- A.
X pripada klasi j, ako je g_j_(x) > g_i_(x), za svako i=1,2..,C, i razlicito od j
- B.
X pripada klasi j, ako je g_j_(x) = g_i_(x), za svako i=1,2..,C, i razlicito od j
- C.
X pripada klasi j, ako je g_j_(x) < g_i_(x), za svako i=1,2..,C, i razlicito od j
Correct Answer
A. X pripada klasi j, ako je g_j_(x) > g_i_(x), za svako i=1,2..,C, i razlicito od jExplanation
The correct answer states that x belongs to class j if g_j_(x) is greater than g_i_(x) for all i=1,2..,C, and different from j. This means that the discriminant function g_j_(x) for class j should have a higher value than the discriminant functions g_i_(x) for all other classes i, indicating that x is most likely to belong to class j.Rate this question:
-
- 30.
Kada je neka mera razlicitosti jednaka 0, to znaci:
- A.
Das u objekti tipicni za dati skup podataka
- B.
Da su objekti maksimalno razliciti
- C.
Da su objekti identicni
Correct Answer
A. Das u objekti tipicni za dati skup podatakaExplanation
When a measure of diversity is equal to 0, it means that the objects are typical for the given dataset.Rate this question:
-
- 31.
Hijerarhijsko klasterovanje formira:
- A.
Skup disjunktnih klastera
- B.
Skup ugnezdenih klastera organizovanih u obliku drveta
- C.
Skup preklapajucih klastera
Correct Answer
B. Skup ugnezdenih klastera organizovanih u obliku drvetaExplanation
Hijerarhijsko klasterovanje formira skup ugnezdenih klastera organizovanih u obliku drveta. This means that the clusters are organized in a hierarchical structure, where each cluster is nested within another cluster, forming a tree-like structure. This allows for a clear understanding of the relationships and similarities between different clusters, as well as the ability to easily navigate and analyze the data.Rate this question:
-
- 32.
Ako je za dati sistem klasifikacije ustanovljena minimalna Bajesova greska odlucivanja, na koji nacin mozemo redizajnirati sistem, tako da se njegova greska smanji ispod ove granice:
- A.
Promenom pravila odlucivanja
- B.
Promenom hardversko softverske realizacije sistema
- C.
Promenom obelezja
Correct Answer
A. Promenom pravila odlucivanjaExplanation
By changing the decision rules, we can redesign the system in a way that reduces its error below the established minimum Bayesian decision error. This means that by modifying the criteria or conditions used to make decisions within the classification system, we can improve its accuracy and reduce the likelihood of making incorrect classifications.Rate this question:
-
- 33.
Klasterovanje se moze posmatrati i na grafovima. U takvoj postavci problema, cvorovima grafa odgovaraju: ????
- A.
Mera razlicitosti izmedju podataka
- B.
Pojedinacni atributi
- C.
Mera slicnosti izmedju podataka
- D.
Podaci
Correct Answer
C. Mera slicnosti izmedju podatakaExplanation
In the given statement, it is mentioned that clustering can also be observed on graphs. In this setup, the nodes of the graph correspond to the measure of similarity between the data points. Therefore, the correct answer is "Mera slicnosti izmedju podataka" which translates to "Measure of similarity between data points".Rate this question:
-
- 34.
Skup reci u nekom dokumentu je:
- A.
Diskretan atribut
- B.
Kontinualan atribut
- C.
Niti kontinualan, niti diskretan
Correct Answer
A. Diskretan atributExplanation
The given correct answer is "Diskretan atribut". This means that the set of words in a document is a discrete attribute. A discrete attribute is a type of attribute that has a finite or countable number of distinct values. In this case, the set of words in a document can be counted and each word is distinct from the others. Therefore, it can be categorized as a discrete attribute.Rate this question:
-
- 35.
Izborom reprezentativnog uzorka se:
- A.
Broj podataka ostaje isti
- B.
Povecava broj podataka za obradu
- C.
Smanjuje broj podataka za obradu
Correct Answer
C. Smanjuje broj podataka za obraduExplanation
By selecting a representative sample, the number of data to be processed is reduced. This is because a representative sample is a smaller subset of the entire population or dataset that accurately represents the characteristics and diversity of the whole. By analyzing this smaller sample, researchers can make inferences and draw conclusions about the larger population, saving time and resources compared to processing the entire dataset. Therefore, the correct answer is that selecting a representative sample reduces the number of data to be processed.Rate this question:
-
- 36.
U okviru procesa primene nauke o podacima u resavanju zadatog problema, tacni su sledeci stavovi: ??
- A.
Podatke nije potrebno prociscavati
- B.
Ako podaci nemaju jedinstvenu prirodu (tip), nuzno se odbacuju
- C.
Izbor algoritama masinskog ucenja je nezavisan od ciljeva postavljenih u okviru problema koji se resava
- D.
Identifikovati relevantne izvore podataka
- E.
Nad podacima je neophodno izvrsiti odgovarajuce transformacije u cilju lakse primene algoritama masinskog ucenja
- F.
Prediktivna moc podataka nije bitna
Correct Answer(s)
D. Identifikovati relevantne izvore podataka
E. Nad podacima je neophodno izvrsiti odgovarajuce transformacije u cilju lakse primene algoritama masinskog ucenja
F. Prediktivna moc podataka nije bitnaExplanation
In the process of applying data science to solve a given problem, it is important to identify relevant data sources. This means finding the sources that contain the necessary information for the problem at hand. Additionally, it is necessary to perform appropriate transformations on the data in order to facilitate the application of machine learning algorithms. This may involve cleaning, preprocessing, and formatting the data. Lastly, the predictive power of the data is not crucial in this context, suggesting that the focus is on using the data to gain insights and solve the problem, rather than solely predicting outcomes.Rate this question:
-
- 37.
Kod asimetricnih atributa su jedino bitne:
- A.
Nulte vrednosti atributa
- B.
Negativne vrednosti atributa
- C.
Ne nulte vrednosti atributa
- D.
Pozitivne vrednosti atributa
Correct Answer
C. Ne nulte vrednosti atributaExplanation
The correct answer is "Non-zero attribute values." This means that only the attribute values that are not equal to zero are important in this context.Rate this question:
-
- 38.
Da li je temperatura:
- A.
Niti je diskretan, niti kontinualan
- B.
Diskretan atribut
- C.
Kontinualan atribut
Correct Answer
C. Kontinualan atributExplanation
The given question is asking about the nature of temperature. The answer "Kontinualan atribut" suggests that temperature is a continuous attribute. This means that temperature can take on any value within a certain range and can be measured with precision. It is not limited to specific discrete values or categories.Rate this question:
-
- 39.
Bajesovo pravilo minimalne greske odlucivanja, kada poznajemo samo apriorne verovatnoce klasa glasi:
- A.
Odlucujemo se za klasu omega_i, ukoliko je P(omega_i) > P(omega_j), za svako j, j razlicito od i
- B.
Odlucujemo se za klasu omega_i, ukoliko je P(omega_i) = P(omega_j), za svako j, j razlicito od i
- C.
Odlucujemo se za klasu omega_i, ukoliko je P(omega_i) < P(omega_j), za svako j, j razlicito od i
Correct Answer
A. Odlucujemo se za klasu omega_i, ukoliko je P(omega_i) > P(omega_j), za svako j, j razlicito od iExplanation
The correct answer states that we choose the class omega_i if the probability P(omega_i) is greater than the probability P(omega_j) for every j, where j is different from i. This means that we select the class with the highest probability among all the classes.Rate this question:
-
- 40.
Ako je mera razlicitosti izmedju dva vektora obelezja normirana na interval (0, 1), kako bi se opisala cinjenica da je mera razlicitosti izmedju dva vektora obelezja jednaka 1:
- A.
Ne vazi nijedan od gornjih stavova
- B.
Ta dva vektora obelezja su maksimlano razlicita
- C.
Ta dva vektora obelezja su identicna
Correct Answer
B. Ta dva vektora obelezja su maksimlano razlicitaExplanation
The given answer suggests that if the measure of dissimilarity between two feature vectors is equal to 1, it means that the two feature vectors are maximally different from each other.Rate this question:
-
- 41.
Endogene metode redukcije dimenzija podataka:
- A.
Nemaju nikakve veze sa informacijama sadrzanim u podacima
- B.
Maksimizuju diskriminacionu informaciju unutar podataka
- C.
Maksimizuju kolicinu informacija o podacima kao celini
Correct Answer
C. Maksimizuju kolicinu informacija o podacima kao celiniExplanation
Endogene metode redukcije dimenzija podataka se odnose na maksimizaciju količine informacija o podacima kao celini. Ove metode se koriste kako bi se smanjio broj varijabli u skupu podataka, ali istovremeno zadržali što više informacija o podacima. Cilj je identifikovati najvažnije karakteristike podataka i smanjiti dimenzionalnost, ali tako da se ne izgubi bitna informacija. Ove metode mogu biti korisne u analizi podataka i modeliranju, jer omogućavaju efikasniju obradu podataka i bolje razumevanje njihovih karakteristika.Rate this question:
-
- 42.
Ako su mere slicnosti izmedju vektora obelezja normirane na interval (0,1), cemu najbolje odgovara cinjenica da je mera slicnosti izmedju dva vektora obelezja jenaka 0:
- A.
Ta dva vektora su maksimalno razlicita
- B.
Nijedan od gornjih stavova
- C.
Ta dva vektora su identicna
Correct Answer
A. Ta dva vektora su maksimalno razlicitaExplanation
The correct answer suggests that when the similarity measure between two feature vectors is equal to 0, it indicates that the two vectors are maximally different. This means that there is no similarity or overlap between the features of the two vectors.Rate this question:
-
- 43.
Ako je g(X)=g_1_(X)-g_2_(X), gde su g_1_(X) i g_2_(X), diskiminacione funkcije za klasu 1 i klasu 2, respektivno, tada se odlucujemo za klasu 1, ako je ispunjeno:
- A.
G(X)=1
- B.
G(X)<0
- C.
G(X)>0
Correct Answer
C. G(X)>0Explanation
The correct answer is "g(X)>0" because if the value of g(X) is greater than 0, it means that the discriminant function for class 1 is larger than the discriminant function for class 2. This indicates that the input X belongs to class 1, and therefore, we decide to classify it as class 1.Rate this question:
-
- 44.
Dobar reprezentativni uzorak ima:
- A.
Aproksimativno iste osobine kao i originalni podaci
- B.
Znacajno se razlikuje po osobinama od originalnih podataka
- C.
Osobine reprezentativnog uzorka nemaju nikakve veze sa osobinama originalnih podataka
Correct Answer
A. Aproksimativno iste osobine kao i originalni podaciExplanation
A good representative sample has approximately the same characteristics as the original data. This means that the sample accurately reflects the population from which it is drawn, allowing for valid inferences and generalizations to be made about the population based on the sample.Rate this question:
-
- 45.
Sekvencijalni podaci se razlikuju po tome sto:
- A.
Redosled nije bitan
- B.
Ne manjeju znacenje ako se permutuju
- C.
Redosled je bitan
Correct Answer
C. Redosled je bitanExplanation
The correct answer is "Redosled je bitan." This means that the order of sequential data is important and changing the order can alter the meaning or significance of the data.Rate this question:
-
- 46.
Koeficijent korelacije meri:
- A.
Trigonometrijsku zavisnost
- B.
Linearnu zavisnost
- C.
Nelinearnu zavisnost
Correct Answer
B. Linearnu zavisnostExplanation
The correct answer is "Linearnu zavisnost" which translates to "Linear dependence" in English. The coefficient of correlation measures the strength and direction of the linear relationship between two variables. It indicates how closely the data points in a scatter plot cluster around a straight line. A high correlation coefficient indicates a strong linear relationship, while a low correlation coefficient suggests a weak or no linear relationship. Therefore, the coefficient of correlation measures linear dependence between variables.Rate this question:
-
- 47.
Redukcija dimenzija podataka po pravilu:
- A.
Smanjuje potrebu za dugackim obucavajucim skupovima
- B.
Povecava potrebu za dugackim obucavajucim skupovima
- C.
Olaksava vizualizaciju
- D.
Omogucava bolji rad algoritama masinskog ucenja
- E.
Ne utice na rad vecine algoritama masinskog ucenja
- F.
Otezava vizualizaciju
Correct Answer(s)
A. Smanjuje potrebu za dugackim obucavajucim skupovima
C. Olaksava vizualizaciju
D. Omogucava bolji rad algoritama masinskog ucenjaExplanation
Dimensionality reduction reduces the need for long training sets by reducing the number of features or variables in the data. This makes it easier to visualize the data as it reduces the complexity and allows for easier interpretation. Additionally, dimensionality reduction enables better performance of machine learning algorithms as it reduces noise and redundancy in the data, leading to improved accuracy and efficiency.Rate this question:
-
- 48.
Hijerarhijsko klasterovanje se vizualizuje u oblliku:
- A.
Dendograma
- B.
Datagrama
- C.
Radiograma
Correct Answer
A. DendogramaExplanation
Hijerarhijsko klasterovanje se vizualizuje u obliku dendograma. Dendogram je grafički prikaz hierarhijske strukture klastera, koji se sastoji od spojenih linija koje predstavljaju sličnost između klastera. Na x-osi se nalaze objekti koji se klasteruju, dok se na y-osi nalazi mera udaljenosti između klastera. Dendogram omogućava vizuelno prikazivanje procesa klasterovanja i identifikaciju grupa objekata koje su slične jedna drugoj.Rate this question:
-
- 49.
Za podatke u Euklidskom prostoru, kao mera kvaliteta klasterovanja se koristi kriterijum:
- A.
Proizvod kvadrata gresaka
- B.
Zbir kvadrata gresaka
- C.
Razlika kvadrata gresaka
Correct Answer
C. Razlika kvadrata gresakaExplanation
In the context of data clustering in Euclidean space, the measure of quality used is the criterion of the difference of squared errors. This means that the quality of the clustering is evaluated based on the difference between the actual values and the predicted values, with the errors being squared to emphasize the importance of larger errors. The larger the difference of squared errors, the worse the clustering quality, as it indicates a larger discrepancy between the predicted and actual values. Therefore, this criterion is used to assess the effectiveness of clustering algorithms in minimizing errors.Rate this question:
-
- 50.
Da li optimalno Bajesovo odlucivanje u teorijskom smislu obezbedjuje minimalnu gresku odlucivanja: ???
- A.
Ne
- B.
Nekad da, nekad ne
- C.
Da
Correct Answer
C. DaExplanation
Bajesovo odlučivanje u teorijskom smislu obezbeđuje minimalnu grešku odlučivanja. Bajesova teorija se zasniva na Bayesovom teoremu koji koristi statističke metode za donošenje odluka na osnovu raspoloživih podataka. Ova metoda uzima u obzir verovatnoću pojedinih događaja i koristi ih za procenu najbolje odluke. Kada se optimalno primeni, Bajesovo odlučivanje može minimizirati greške odlučivanja i obezbediti najbolji mogući rezultat.Rate this question:
-
Back to top