QM

Question 1

R^2

Answer 1

Mide la varianza explicada por el modelo

Question 2

AIC

Answer 2

Pediction o forecasting (particular models)

Question 3

BIC

Answer 3

Goodness of fit (Parsimonious - General model)

Question 4

Significancia

Answer 4

|statistic|> Critical Value
p-value<alpha
Reject de null

Question 5

Conditional Heteroskedastic

Answer 5

Inflated statistics (Error Type 1), SE underestimated, biased coefficients, underestimated p-values

Question 6

Serrial Correlation

Answer 6

Inflated statistics (Error Type 1), SE underestimated, negative sc implica inconsistent coefficients

Question 7

Multicolinearity

Answer 7

Inflated SE, se corrige ampliando la muestra, excluyendo más variables y/o usando proxys

Question 8

Logistic Regression

Answer 8

Outcome discreto

Question 9

Dick fuller Test

Answer 9

Unit root, donde H0: g=0 y Ha: g<0, g=b1-1, fail to rejcect implica Unit root (RW)

Question 10

Seasonality

Answer 10

Significancia en un periodo o autocorrelation positiva en residuals

Question 11

PCA

Answer 11

Dimension Reduction of highly correlated features (continous unsupervised)

Question 12

Regression

Answer 12

Prediction (continous supervised)

Question 13

Regression or classification complex non-linear data

Answer 13

CART, RF, NN

Question 14

Regression no complex non-linear data

Answer 14

Penalized Regression

Question 15

Classification Labeled Data

Answer 15

Discreto supervised

Question 16

Classification no complex non-linear data

Answer 16

KNN, SVM

Question 17

Classification no Labeled Data

Answer 17

Clustering Continous Unsupervised

Question 18

Clustering no complex non-linear data (#categories)

Answer 18

K-means

Question 19

Clustering no complex non-linear data ( No #categories)

Answer 19

Hierichal Clustering

Question 20

Clustering complex non-linear data

Answer 20

NN

Question 21

Overfitting

Answer 21

Variance error

Question 22

Underfitting

Answer 22

Bias Error

Question 23

SVM

Answer 23

Outlier detection, target variable binary, no defined hyperparameter

Question 24

KNN

Answer 24

Clasifica nuevas observaciones encontrando similitudes en las existentes (k puntos más cercanos - #clusters), sensitive to local outliers

Question 25

Ensamble Learning and RF

Answer 25

Prediction of a group of models combination (prevent overfitting)

Question 26

Bootsrap Aggregating

Answer 26

Original training data set is used to generate n new training data sets

Question 27

K-mean clustering

Answer 27

Centroid -> Cluster iteration until no observation is reassigned (hyperparameter = # clusters)

Question 28

Hierachical Clustering

Answer 28

Agglomerate clustering (cluster de cluster)

Question 29

Qué es CH

Answer 29

Var(errors) no constant

Question 30

Medición CH

Answer 30

BP Test, T-test: Chi-cuadrada =nR^2 > Chi-cuadrada alpha implica CH

Question 31

Qué es SC

Answer 31

Correlation of errors

Question 32

Medición SC

Answer 32

DW < Critical Value implica SC BG > Critical Value implica SC

Question 33

Qué es MC

Answer 33

Correlation entre IVs

Question 34

Medición MC

Answer 34

VIF > 10, high R^2 y P-value=0.0000 implica MC

Question 35

No residual autocorrelation

Answer 35

No significancia en todos los coeficientes o alguna autocorrelation difiere mucho de cero implica no reject de null que impica no SC

Question 36

Testing Joint coefficientes hipótesis

Answer 36

H0: bi = 0 oara todo i (unrestricted model bi son las variables quitadas)

Question 37

Unconditional heteroskedasticity

Answer 37

Errores no correlated

Question 38

Autocorrelation T-stati

Answer 38

autocorrelationi/SE, no se usa DW

Question 39

Random Walk

Answer 39

No cov stationary, b0=0, b1=1

Question 40

First differencing

Answer 40

Convertir a cov stationary b0=0 y b1=0

Question 41

RW drift

Answer 41

b0<>0, b1=0

Question 42

Multiple TS (both reject DF)

Answer 42

No unit root

Question 43

Multple TS (Only one reject)

Answer 43

Error no cov stationary que implica SE inconsistent

Question 44

Multiple TS (Both unit root)

Answer 44

Test for cointegration with Engle-Granger

Question 45

No cointegration

Answer 45

Error no cov stationary que implica SE inconsistent

Question 46

Cointegration

Answer 46

OK! Long term economic relation exist

Question 47

Prevent overfitting

Answer 47

Cross validation (form the same domain as training data)

Question 48

K-fold cross validatrion

Answer 48

Data shuffled randomly divided into k sub samples (a mayor mejor con target variable specified)

Question 49

Manage overfitting

Answer 49

trade off between cost and complexity

Question 50

Neutral Networks (NN)

Answer 50

Input data (0,1) and weights chosen to minimize loss

Question 51

Deep learning nets

Answer 51

Neutral Networks with many hidden layers (at least 3, typically >20)

Question 52

Reinforcement Learning

Answer 52

Unsupervised: learns by testing new actions through millions of trials and errors

Question 53

Data sets

Answer 53

Training, validation (tuning the model), test

Question 54

Characteristics of data sets

Answer 54

Volume (quality of data), variety (structured and unstructured), velocity y veracity

Question 55

Process of data sets

Answer 55

Task-collect data (curation)-preparation (wrangling)-exploration-training

Question 56

invalid data

Answer 56

outside of meaning full range

Question 57

Inaccurate data

Answer 57

no true values

Question 58

inconsitent data

Answer 58

NY in Canada

Question 59

non uniform data

Answer 59

jan-15 vs 15/01

Question 60

extraction of data

Answer 60

new variable created from a current one

Question 61

aggregation data

Answer 61

more than 1 variable aggregated into one

Question 62

selection of data

Answer 62

columns that can be eliminated

Question 63

Conversion of data

Answer 63

Change the type

Question 64

Trimming

Answer 64

delete 1% de top y bottom

Question 65

winsorization

Answer 65

replace outliers with min/max values

Question 66

Scaling normalization

Answer 66

(Xi - min)/(max-min)

Question 67

scaling standarization

Answer 67

(Xi - m)/sigma

Question 68

Text cleansing

Answer 68

Remove caracteres innecesarios

Question 69

Tokenization

Answer 69

Splitting text into tokens

Question 70

Normalizing data

Answer 70

Lowercase o preposiciones

Question 71

Stemming

Answer 71

analyzed to analyze

Question 72

Lemmatization

Answer 72

anlayz to analyze

Question 73

One hot encoding

Answer 73

decomposing de multiples a un solo valor

Question 74

Exploratory analisis

Answer 74

.Charts correspondientes (structured y unstructured)

Question 75

Feature selction

Answer 75

Eliminate uneed features

Question 76

Feature engineering

Answer 76

creating a new feature and decomposing

Question 77

Feature selection methods

Answer 77

Frecuency ratio (frequent underfitting vs sparse overfitting), Chi-square test, mutual information (distribución de un token a una clase)

Question 78

Model training

Answer 78

Method selection, clssification and size of data

Question 79

Precission

Answer 79

TP/(TP+FP), no caer en error type 1

Question 80

Recall

Answer 80

TP/(TP+FN), no caer en error type 2