Please enable JavaScript.
Coggle requires JavaScript to display documents.
modeling users attitudes bran (OBSTACLE FORCES (inference of share of…
modeling users attitudes bran
approach for explaining the data
vector autoregression
use granger causality
statiistical hypotheses for determinig whether one time series is useful in forecasting another
lagged values of dependent variables
test case model
independent variablesof the experiment
dependent variables
base case autoregressive model
what do we want to test
second lagged independent variables
redo 3 steps above in reverse direction. check it between test and p-value.and see whcih variables have more granger casualiy
what is the optimum value of lag?
what values of lags should be tested?
calculation of errors for base case and test case
F-test in which cases?
T- test in which cases?
for all the variables, check adding which variable as second variable would cause siginificantly lower error.
data
advertise data
sales data
text mining of news
modelling the correlation between sales and advertisements
share of market
share of expenditures
volume of coverage of news
chose the appropriate number of previous values
interpolate data by a technique called local regression
weight of the tone of the news
having them all for estimation
ex
back logof everything we need to know or to have
how too build a statistical hypotheses
OBSTACLE FORCES
inference of share of voice
inference of news tonality
only algorithms with high accuracy low error of prediction/
what other factors ( like time frame of recent data) have been considered in measuring granger causality.
how to make samples introduced as granger values, what and where to find them/
GOAL :check:
find a model that , can show that sales and advertisement has vinterdependency and find its correlation measure
correlation between news and advertisement
ambiguities about its measurement
time series , what time series to be considered?
finding structural differences in experiment that could make THIS APPROACH inefficient
polls and news reflect all aspects of customer expectations. what might happpen when some aspects are missed in data?
uses of this correlation
f
concept proof problems
showing granger causality
index to show if two things can forecast each other and which one has more effect