Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
96 Cards in this Set
- Front
- Back
|
Computer model |
a set of mathematical relationships and logical assumptions implemented in a computer |
|
|
what are spreadsheet models the most convenient and useful for |
business people to make decision alternatives before having to choose a specific plan for implementation |
|
|
what do we consider to make good decisions |
multiple criteria of varying importance and chose the best course of action |
|
|
what do business analytics use to solve business problems |
data, computers, stats, and mathematics |
|
|
what is business analytics also referred to as |
operations research, management science, decision science |
|
|
business analytics |
the scientific discipline devoted to the analysis and solution of complex management decisions |
|
|
decision technology |
collected of computer based methods and tools for building manipulating and solving models |
|
|
business analytics tools |
1. excel 2. treeplan 3. solver 4. analytic solver platform |
|
|
mental model |
visualize the outcome |
|
|
visual model |
blueprints, maps |
|
|
physical or scale models |
prototypes of final design |
|
|
mathematical model |
mathematical relationships are used to describe a decision problem -often spreadsheets are the tool for building mathematical models |
|
|
benefits of modeling |
1. simplification 2. represents relevent characteristics 3. less expensive 4. deliver need info on a timelier basis 5. examine things that would be impossible to do in reality 6. insight and understanding |
|
|
spreadsheet functions are used because |
1. faster 2. more accurate 3. scalable 4. flexible |
|
|
quantitative modeling |
much of the art of management is taking the vast quantities of data and other inputs and making sense of it, and using it to improve the business |
|
|
deterministic mathematical model |
known and well defined: stats, linear and integer programming |
|
|
probabilistic mathematical models |
predictive: forcasting descriptive: inputs are unknown |
|
|
problem solving process |
1. define the problem 2. model the problem 3. solve the model 4. communicate the results |
|
|
benefits of using a model |
1. cost 2. speed 3. flexibility |
|
|
success |
converting model results into business insights and using those insights to improve the business |
|
|
technical success |
the model works |
|
|
organizational success |
the model is accepted and used |
|
|
end user development |
building models yourself |
|
|
user interface |
particularly important when others are using the model |
|
|
documentation |
written records of the model |
|
|
user documentation |
how to run/use the model |
|
|
model documentation |
purpose, assumptions, output format |
|
|
programming documentation |
coding, cell definitions, comments |
|
|
hierarchy of modelling skills |
1. numeracy and logic skills 2. basic modelling skills 3. advanced modelling skills 4. management science/business analytics tools and applications |
|
|
two errors in human judgement that can impact decisions |
1. anchoring 2. framing |
|
|
anchoring |
depends on your starting point |
|
|
framing effects |
the decision is based on how the question is asked, the decision makers perception of risk and/or how it would impact the decision maker personally |
|
|
absolute references |
used in excel formulas to facilitate copying if there is a $ it is absolute and will not change when copying or else it is relative |
|
|
IF |
(condition, result if true, result if false) |
|
|
Vlookup |
purpose: to look up data automatically benefits:time savings, accuracy |
|
|
function to use for highest score |
max |
|
|
function to use to find second highest score |
LARGE |
|
|
to find lowest score |
min |
|
|
to find second lowest |
small |
|
|
Countif |
useful to count the number of items that fall in a certain category |
|
|
pivot table |
the ability to rearrange data along different dimensions very quickly |
|
|
best practices for spreadsheet modelling |
1. a clear, logical layout to the model 2. seperation of different parts of the model 3. clear headings 4. use of range names 5. liberal use of bolduse, italics or larger font size 6. use of cell comments 7. use of text boxes |
|
|
mathematical programming |
A technique for allocating limited resources most effectively when there are competing demands for these resources. referred to as optimization |
|
|
simplest form of mathematical programming |
linear programming |
|
|
linear programming |
a deterministic technique because all of the input data and parameters are known with certainty |
|
|
some applications of mathematical optimization |
1. product mix and manufacturing 2. routing and logistics 3. financial planning |
|
|
characteristics of optimization |
1. one or more decisions must be made 2. there is come goal or objective that the decision maker is trying to achieve, most commonly 3. there are restrictions or constraints that are placed on the alternatives available to the decision maker |
|
|
basic assumptions of an LP model |
1. certainty: all parameters are known and are constant 2. proportionality: the objective function and constraints are linear (proportional) 3. additivity: the total of all activities is the sum of the individual activities 4. divisibility: solutions need not be whole numbers |
|
|
steps in formulating LP models |
1. understand the problem 2. identify decision variables 3. state the objective function 4. state the constraints |
|
|
common types of constraints |
1. availability of resources 2. availability of markets 3. composition 4. proportional relationships |
|
|
feasible soltion |
a solution that satisfies all the constraints |
|
|
infeasible solution |
a solution that does not satisfy one or more of the constraints |
|
|
optimal solution |
the feasible solution that results in the maximum/minimum objective is called the |
|
|
alternate optimal solutions |
- sometimes there may be more than one "best solution" thus we have alternate optimal solutions -solver doesn't indicate whether there are alternate optimal solutions -this is a rare case |
|
|
redundant constraints |
- sometimes a constraint may not impact the feasible region because the existence of all the other constraints mean that is it always satisfied. therefore the constraint is not strictly necessary
|
|
|
unbounded |
(no limit) solver has recognized that the solution is an infinite value - likely forgot to indicate a constraint |
|
|
infeasability |
solver has determined that there is no set of values for the changing cells that satisfies the constraints; you need to examine the constraints to identify the conflict - often a sign reversal |
|
|
nonlinearity |
you have selected the simplex lp option but solver determines that your model is not linear - check constraints or the target cell for formulas that are non linear
|
|
|
solver |
Computer implementation of some spophisticated mathematical algorithms that search cleverly for viable cell values that yield good target cell values |
|
|
binding constraints |
where all of a resource that is available is used (total=available)
have zero slack in the optimal solution |
|
|
shadow price |
indicates the amount by which the objective function value changes given a unit increase in the RHS call of the constant - it is the value of one additional unit of a scarce resource. |
|
|
a binding constraint with a non zero shadow price shows what |
indicates a scarce resource |
|
|
change in target cell = |
change in resource X shadow price |
|
|
reduced cost |
is the amount that profit would have to change by to make it worthwhile to make product at the optimal solution |
|
|
is forecasting always right? |
no basically always wrong |
|
|
which type of forecasts are the most accurate |
short range over long range |
|
|
time series |
a set of observations on a quantitive variable collected over time |
|
|
quantitative methods of forecasting |
1. time series forecasting - relies on past data to predict the future 2. causal methods - regression, leading indicators, econometric models - includes use of external factors
|
|
|
qualitative forecasting |
when no historical data is available or in an environment of extreme change |
|
|
qualitative approaches to forecasting |
1. delphi and nominal group techniques 2. expert judgement 3. intuitive approaches 4. product life cycle curve |
|
|
selection criteria for which forecasting method to use |
1. forecast horizon 2. required accuracy 3. data availability 4. resources available to make forecasts |
|
|
accuracy of forecasts |
1. forecasting involves error 2. forecasts should include a measure of error 3. family forecasts are more accurate than individual item forecasts 4. short range forecast are more accurate than long range |
|
|
Characteristics of a good forecast |
1. quality - 2 measure a) accuracy: size of forecast errors b) bias: were predictions consistently high or low? 2. cost 3. responsiveness: should reflect changes in market conditions quickly 4. timeliness: should be available at the time decisions have to be made 5. simplicity: should be easy to understand
|
|
|
stationary time series |
there are no significant upward or downward trend in data over time |
|
|
how would you forecast if you had little or no historical data |
1. intuition 2. look at similar products 3. if enough time do a market survey |
|
|
underlying assumption for using time series analysis |
what happened in the past is a good indication of the future |
|
|
what are smoothing methods for data that is relatively stable |
1. simple moving average 2. weighted moving average 3. exponential smoothing |
|
|
Ft |
the forecast of an unknown value for some period t in the future |
|
|
Yt |
the actual observed known value for some period t in the past |
|
|
et |
the error between the actual and the forecast value |
|
|
formula for et |
et= Yt- Ft |
|
|
simple moving average |
the average in the last "n" previous observations in the series. when you get new data you update the forecast with the last "n" time periods |
|
|
weighted moving average |
assigning different weights to the data - sum of the weights must equal 1 with generally higher weights for more recent data |
|
|
when does WMA work better then SMA |
when there is a demand trend however it still lags behind in the trend significantly |
|
|
simple exponential smoothing |
average technique for stationary data that allows weights to be assigned to past data. carries along all historic demand data but weights recent demand more heavily |
|
|
what does a do? |
it is a smoothing parameter that weights the relative influence of recent data and older data |
|
|
what does higher smoothing parameter mean |
faster response to actual demand but more fluctuation in forecast |
|
|
performance measures |
1. bias (average error) 2. MAD ( mean absolute deviation) 3. RMSE (root mean square error) 4. MAPE ( mean absolute percent error: this is good for comparing different data sets) |
|
|
steps to comparing and selecting forecasting methods |
1. identify alternative forecast methods and parameters 2. apply to historical data and determine the errors of each alternative 3. select the best forecasting method (lowest error values) to use in the future |
|
|
trend |
upward or downward movement in a time series - employ a base + trend tool |
|
|
seasonal |
cyclical or repeating pattern -employ a base*cyclical index tool |
|
|
trend projection |
fits a trend line to a series of historical data points and then projects the line into the future for medium to long-range forecasts |
|
|
holts linear approach method |
forecast = base + k (trend) where k is the number of periods into the future that we are forecasting |
|
|
Cyclical index |
forecast= base * cyclical index |
|
|
steps to forecasting with cyclical pattern |
1. using historical data, determine average demand for a period 2. using historical data, calculate cyclical index for each period 3. apply each period's cyclical index to expected future demand |
|
|
what might cause outliers |
1. data collection/recording error 2. situations causing extreme demand |
