Fichas
Midterm Flashcards
Fichas del conjunto (100)
- 1Common elements of decision making problemsDecision alternatives
States of nature
Payoffs - 2Decision AlternativeSet of various choices or options available to decision maker
- 3States of natureOccurrence of nature that can affect the outcome of the decision
Decision make has little to no control over - 4PayoffsBenefits or rewards that result from selecting particular decision alternative
Net gains or losses - 5How is a payoff table structuredDecision alternatives/options in rows
Possible outcomes in columns - 6Decision criteriaFactors that are important to the decision maker
- 7Which decision making options are non-probabalisticMaximax
Maximin
Minimax regret - 8Which decision making options are probabalisticExpected Monetary Value (EMV)
- 9MaximaxMax of max
Maximizes largest gain
Optimistic
Select action where best (maximum) gain is better than the best gain of all other courses of action possible in given circumstances. - 10MaximinMax of Mins
Maximizes the smallest gain
Pessimistic/Conservative.
Select action which where Worst (maximum) loss is better than the least (minimum) loss of all other courses of action possible in given circumstances. - 11Minimax RegretFinds max loss for each decision and chooses the decision with the smallest max loss
Minimum of the maximum regrets/losses - 12Regret=best payoff-payoff recieved
- 13Steps to constructing a minimax regret table1. Examine the outcome
2. Put all results into opportunity loss table showing loss values for each decision to outcome pairing
3. Find max loss regret for each decision
4. Best decision is min of max losses - 14EMV-Expected Monetary ValueWeighted average of possible payoffs for a decision, weighted by the probabilities of the outcomes
- 15EMV calculation=sumproduct(array 1,array2)
- 16Where do you add probabilities on a decision TreesAt outcomes
- 17Where are EMV's calculated on a decision treeAt chance nodes
- 18EVPI-Expected value of perfect informationThe difference between the payoff that would occur if the decision maker knew the outcomes and the payoff from best decision when there is no information about the occurrence of the states of nature
=EMV with free perfect information-EMV wo perfect information
=The most you would be willing to pay for perfect information - 19EVSIDifference between payoff if decision maker had sample info on likelihood of which states of nature would occur and expected monetary payoff from best decision alternative where there is no info
=EMV with sample info-EMV without - 20Conditional ProbabilityProbability of an event or outcome given that another event or outcome has occured
Probability of A given the occurence of b
P(a/b) - 21Contingency tableLists probability of event A event B
Categories in the rows and columns must be mutually exclusive - 22Mutually exclusiveCannot occur at the same time
- 234 components to time series data1. Trend
2. Seasonal
3. Cyclical
4. Random - 24TrendLong run increase or decrease over time
Data taken over a long period of time - 25SeasonalShort term regular patterns
Observed within a year
i.e. monthly, quarterly - 26CyclicalLong term wave like patterns
Measured peak to peak - 27RandomUnpredictable residual fluctuations due to random variations of nature or accidents
"noise" in time series - 28Stationary dataNo upward or downward trend
Most business and economic data is not stationary - 29Two methods of forecastingEconometric and extrapolation
- 30EconometricAKA Casual or explanatory
Uses regression to forecast a time series variable by using other explanatory time series variables
-Regression - 31ExtrapolationUses past data of time series variables to forecast future values
-Avg
-Exponential Smoothing
-Trend models - 32Components to look at when forecastingPatterns in previous data sets
- 33If data is stationary, which methods would you chooseMoving average, weighted moving average, simple exponential smoothing
- 34If data is trending with no seasonal effects, which method would you useHolt's
- 35If trending with seasonal effects, which method would you useHolt's Winters
- 36When can you use the averaging methodWhen sales are not trending
Treats all data as equally important
Sum of data /#of data points - 37When can you use the moving average method-When you want to control the amount of past data used
-Span of (k) is the amount of past data you will use
-Treats all data as equally important
i.e. k=5, in step six you average data points 1 through 5
=Sum of data/span - 38Difference between a small and large span(k)Small k ignores most of our history
Large k does not give much weight to recent history and smoothes out fluctuations in data - 39Weighted moving averageTreats some data as more important than other data
Weight(w)1 is most important, weight 3(w3) is less
=data*weight+data*weight+data*weight - 40How to determine which method is the most accurateMethod with lowest RMSE
- 41Simple exponential smoothing-Form of weighted average
-Considers ALL data prior to time being forecasted
-Most recent data weighted at the most
-Forecast for first period without actual data is forecast for all future periods - 42Methods for Stationary data-Average
-Moving Average
-Weighted Moving Average
-Simple Exponential Smoothing - 43What does L stand forLevel
- 44What does Y stand forObserved data/
Actual data - 45What does K stand forSpan
How many periods since the last observation - 46What does B stand forB=Beta
Smoothing constant that ranges in value from 0 to 1 - 47Formulas for Holt's Model for TrendInitial Level
Initial Trend
Level
Trend
Forecast - 48What does forecast =Forcast=Level + Trend
- 49What does S stand forSeasonality
- 50How many factors do you have if the sessions of seasonality is measured in months12
S1-S12 - 51How many factors do you have if the sessions of seasonality is measured quarterly4
S1-S4 - 523 steps to Holt's model application1. Initial Conditions
2. Calculate subsequent levels & Trends
3. Forecasts - 53What are included in the initial conditionsInitial Level
Initial Trends - 54Equation for initial levelL2=Y2
Lsep=Ysep
Level of sep=actual data for sep - 55Equation for initial trendT2=Y2-Y1
TSept=Ysept-Yaug
Trend for sept=Actual data for sept-actual data for aug - 56What is Holt's Winter's and what two factors does it take into accountHolt's Winters is a form of exponential smoothing that takes into account trend and seasonality
- 57Formulas for Holt Winter's ModelInitial Level
Initial Trend
Level
Trend
Seasonal Factor
Initial Seasonal Factor
Forecast - 58Components of a good forecastA good forecast will have probabilistically independent errors
The error from one time period has no relationship with the error for the previous time period - 59How to tell which forecasting method is appropriateErrors should not have a pattern and should be evenly distributed above and below 0
If the errors are mostly positive then it's bad - 60What is regressionIt is an explanatory method
It examines the relationships between variables
How a single variable depends on another variable
How the cause relates to the effect - 61Dependent variableDepends on explanatory variable
Goes on the Y axis - 62Explanatory variablesOne or more variables used to explain dependent variable
X axis - 63Purpose of regressionTo understand how the cause relates to the effect
To make predictions - 64Type of data used in regressionCross sectional data
Data gathered from the same period of time from a cross section of the population
Time Series data
One or more variables that are observed at points in time - 65Difference between single and multiple regressionSingle uses a single explanatory variable
(i.e. shoe size based on height)
Multiple uses any number of explanatory variables
(i.e. wage based on role, gender, exp) - 66What do you do before beginning regression analysisCreate a scatterplot
- 67What do scatter plots help you identifyIf there is a linear relationship or trend
- 68What is a correlationa measurement of the strength of the linear relationship
can range between -1 and 1
Only relevant to linear relationships - 69What does a correlation equal to or near zero indicateNo linear relationship
- 70What does a correlation with a magnitude close to 1 indicateA strong linear relationship
- 71What does a positive correlation value between two variables indicateAn upward trending relationship
- 72What does a negative correlation value between two variables indicateA downward trending relationship
- 73What does the coefficient indicateThe coefficient indicates that for every unit increase in x y will change by coeffiencent
- 74What happens to the slope of the line if there is a change in xA change in Y happens when x increases by one unit
- 75Regression equationPredicted Dependent Variable=Coefficient of the Intercept+Coefficient of explanatory variable*explanatory variable
I.e
Predicted Sales=25.123+0.762*promote - 76What good qualities are examined in a scatterplot of residuals produced by a regression analysisGood qualities that are searched for include
small residuals
Residuals scattered around zero
Residuals with no patterns - 77In the regression statistics table, list where to find the following Absolute value of correlation Coefficient of determination Standard error of estimateAbsolute value of correlation=Multiple R
Coefficient of determination=R2
Standard error of estimate=Standard Error - 78What does the coefficient of determination doThe coefficient of determination, or R2 measures the amount of variation in the dependent variable that is explained by the explanatory variable
Always ranges between 0 and 1 - 79What does is mean when R2 is close to zeroVariation in x has no effect on y
- 80Interpret R2=0.452945.29% of sales is explained by promotion
- 81What is a populationData set that includes all of the entities of interest in a study
- 82What is a samplesubset of the population
- 83What functions do you use for normal distributionNorm.dist and Norm.inv
- 84List the functions used for -less than probabilities -greater than probabilities -between probabilities -percentile calculations-less than=norm.dist(z,1)
-greater than=1-norm.dist(z,1)
-between probabilities=norm.dist(zb,1)-norm.dist(za,1)
-Percentile=norm.inv - 85How much of the data falls between 1, 2 and 3 standard deviations1=68%
2=95%
3=99.7% - 86Regression equation in multiple regression if Y=predicted overhead and x=machine hours, production runsPredicted Overheard=0000*machine hours+0000*production runs
- 87Interpret the equation 3997+43.54*Machhrs+883.62*ProdRunsIf the number of production runs is held constant, the overheard cost is expected to increase by $43.54 for each extra machine hour
If the number of machine hours is held constant, the overhead is expected to increase by $883.62 for each extra production run
$3997 is the fixed overhead - 88What happens as you add explanatory variables to a regressionThe R2 goes up
- 89What variable should you examine when using multiple regression and whyExamine Adj R2 as the R2 will always go up as you add variables. Adj R2 does not always go up and can indicate that the variable belongs in the equation
- 90What does a T-stat indicateA t-stat indicates how many standard errors the regression coefficient is away from 0
if b=0 than the x variable has no effect on the y - 91What is the null hypothesis testThe null hypothesis test tests the validity of the regression coefficient. If b=0 than the x variable has no effect on the y. If the null hypothese is less than 0.05% we discard the null hypothesis and assume that the coefficients have some explanatory power
- 92Difference between Tstat and FstatTstat test which of the potential variables are useful in explaining the dependent variable
Fstat test tests all of coefficients as a group for significance
Tests for linear relationship between Y and all independent variables - 93Do you want Fstat to be high or lowHIGH
You want your explained variation to be higher than your unexplained variation - 94Multiple RCorrelation value
between 0-1
higher value indicates better predictability
of dependent variable from independent variable - 95R-SquareCoefficient of determination
Measures the proportion of total variation of outcomes explained by the regression - 96Adjusted R2Adjusts R2 for the number of explanatory terms in a model
Increases only if new explanatory variable improves regression - 97Standard ErrorStandard error of estimate
Measure the accuracy of prediction - 98FtestTests for existence of linear relationship between variables
large f stat means explained variation is large relative to the unexplained variation - 99P ValueProbability of the null hypothesis being true-that none of the explanatory variables explain the variance in the dependent variance
Low P values are good-less than 0.05 (5%) - 100—
Información
- Descripción
- Decision Analysis Fichas sobre Midterm Flashcards, creado por Jennifer Pastega el 19/02/2017.
- Creado por
- Jennifer Pastega
- Visitas
- 2
- Valoración