KKT optimality improvement #165
Description
Hi All, I have three years of retail purchase transaction level data. I keep getting "Warning: The KKT optimality conditions are not both met for the fitted gg spending model", but it does predict the CLV. Does this mean, the CLV predicted is NOT reliable. I do see that P values for S and Beta are high ( not significant). Pls see below.
est.pnbd <- pnbd(clv.data = clv.train)
Starting estimation...
Estimation finished!
summary(est.pnbd)
Pareto NBD Standard Model
Call:
pnbd(clv.data = clv.train)
Fitting period:
Estimation start 2017-12-31
Estimation end 2020-09-06
Estimation length 140.0000 Weeks
Coefficients:
| Estimate | Std. Error | z-val | Pr(>|z|) | |
|---|---|---|---|---|
| r | 0.8835 | 0.1021 | 8.655 | < 0.0000000000000002 *** |
| alpha | 25.604 | 3.3775 | 7.581 | 0.0000000000000344 *** |
| s | 0.5035 | 0.6359 | 0.792 | 0.429 |
| beta | 127.5049 | 220.6421 | 0.578 | 0.563 |
Signif. codes: 0 ‘’ 0.001 ‘’ 0.01 ‘’ 0.05 ‘.’ 0.1 ‘ ’ 1
Optimization info:
LL -6033.2729
AIC 12074.5458
BIC 12090.1998
KKT 1 TRUE
KKT 2 TRUE
fevals 38.0000
Method L-BFGS-B
Used Options:
Correlation FALSE
results <- predict(est.pnbd, continuous.discount.factor = d )
Predicting from 2020-09-07 until (incl.) 2021-01-31 (21.00 Weeks).
Estimating gg model to predict spending...
Starting estimation...
Estimation finished!
Warning: The KKT optimality conditions are not both met for the fitted gg spending mode