gavina
"More data always leads to better forecasts"
conditionally true
More data helps only if it is relevant and high quality
too much data can cause noise and confusion
lead to overfitting
when qualitative outperforms quantitative
when data is a disadvantage
there is no historical data
situation is new or rapidly changing
human inside the text patterns that data cannot
during disruptive innovation
when past trends become misleading
why quantitative forecasts fail
wrong model selection
structural changes
Lita misinterpretation
why I'm certainty cannot be eliminated
The future contains unknown variables
human behavior is unpredictable
external shocks cannot be modeled perfectly
realiability holds when
reliability breaks down when
data is relevant and updated
experts are unbiased and experienced
proper forecasting techniques are used
Data is outdated or incomplete
bias dominates judgment
sudden environmental changes occur
forecasting a new products
step by step approach
1 use qualitative methods
2 analyze similar products data
3 apply quantitative models to estimated figures
4 continuously update forecast as real data comes in
forecast higher than actual results
diagnosis and corrections:
develop basis
estimate operations
regulate forecast
review process
supposed to experts give opposite qualitative forecasts about the same event
check track record of each expert
evaluate logic and consistency
assess bias and motivation
if all forecasts are based on past and present data explain how forecasting can still account for unprecedented events
using scenario analysis
considering possible risks
building flexible models
is forecasting a science or art
science
it relies on systematic methods and data analysis
uses structured processes and models
problems in forecasting a perfectly accurate forecast
Make companies overconfident
reduce flexibility
need to overdependence on prediction
hidden assumption
The future will resemble the past
if false
historical data becomes useless
forecasting models collapse
if a forecast influences decisions and those decisions change the future is the forecast is still valid or does it invalidate itself
forecasts influence decisions
decisions change outcomes
the original forecast may no longer match reality
create your own forecasting model
budget influencing forecast
14 and 2