Forecasting Flashcards
(38 cards)
Correlation
two variable if a change in the value of one is accompanied by a change in value of the other variable
- distance of journey and time takes
- scatter graphs best way of ascertaining
What are the two main ways to estimate a line of best fit for scatter graph data?
- High Low - two points to fill i line equation
- Linear Regression - finds equation for line of best fit, formula provided by
What does the usefulness of using linear regression depend on?
predicting results in future depends on how strong correlation between two variable are
What are the adv.s of linear regression analysis
- definitive line of best fit
- makes efficient use of data
- good results can be obtained with small amounts of data
- many processes are linear and so described well here
Interpolation
- using line of best fit to predict value within two extreme points if the observed range
Extrapolation
- using line of best fit to predict value outside the two extreme points of observed range
- further extrapolate = less accurate predictions
Correlation Coefficient?
r
- measures degree of linear correlation between two variables
- nearer r is to +1 (perfect positive correlation) or -1 (perfect negative correlation) stronger relationship
- indicates strength of linear relationship
formula provided to find
What are the disadvs.s of linear regression?
- assume linearity between x and y
- observations used may be atypical
- historic data used and future patterns may change
- each observation should be independent but in reality success affects future sales
Coefficient of determination?
r^2
- measures proportion of total variation in value of variblae that can be explained by variations in other variable
- denotes strength of linear association between variables
- does not prove cause and effect, just possible link
- value between 0 - 1
f% variation costs can be explained by chgs in output
Time Series
series of figures or values recorded over times (TS)
- analysis of historical data allows observations of how variable has performed
- monthly sales over last five years
- output at factory over a month
Trend
- component of TS
underlying long- term movement over time in values of data recorded (T) - generally expected to be smooth line/curve
- can be found using moving averages
Seasonal Variations
- short term fluctuations in recorded values (SV)
- due to different circumstances that affect results at different times of day/week/year/regularly repeating pattern
Cyclical Variations
- recurring patterns over longer period of time (CV)
- generally not of a fixed nature
- recession/economic growth
Random Variations
- irregular/unpredictable variations due to rare/chance occurrences (RV)
- hurricanes/floods/pandemics
The additive Time Series model
TS = T + SV + CV + RV
only valid if components are independent
in exam:
TS = T + SV
The multiplicative time series model
TS = T x SV x CV x RV
in exam:
TS = T x SV
How can a trend be calculated from a time series?
moving averages
- averages all results of fixed number of periods and relates to the mid-point of overall period
period of the moving averages depends of TS
- quarters 4- period
- 5 day week- 5- period
What is the process of working out the trend using moving averages?
Decide on period amount (3)
take first 3 and average (add together/3)
- this average relates to middle of 3 values
- continue on next 3 and so on
- will have moving averages for all but two extreme values and can view trends
What is the process of working out trend with moving averages for when even number of results?
- mid point won’t relate to single period so take moving average of moving average
- moving total of 4 quarters
- moving average of 4 quarters (added/4)
- mid point if 2 moving averages = trend line of
What do you do with the moving average to work out Seasonal Variation for both models.
Additive- Times Series - Trend (moving average)
They should sum to 0
Multiplicative - Time Series/ Trend
They must equal the number of periods over which seasonality occurs prior to repeat.
- this is better when there is inc/dec in trend
either case adjustments made to ensure sum correctly
What are the advs of time series analysis ?
- trend lines can be reviewed after each period for reliability
- leading to improved forecasts
- relatively easy for non-financial managers to understand
What are the disadvantages of time series analysis?
- further into future = more unreliable
- less data available = less reliable
- pattern of trend/SV can’t be guaranteed to continue
- always danger of random variations
What at index numbers often used for?
- by economists to provide a standardised way of comparing values over time
What does an index measure?
- average change in values, prices or quantities of a group of items
- instead of getting feel of trend with time series can find actual figures
- starts with base year, at index number of 100
