There is an issue on chapter Evaluating Causal Models, in the following paragraph
First, look at the random model (rand_m). It has roughly the same estimated elasticity in each of its partitions. We can already see just by looking at the plot that it won’t help us much with personalisation since it can’t distinguish between the high and low price elasticity days. Next, consider the predictive model, m_pred. That model is actually promising! It manages to construct groups where the elasticity is high and others where the elasticity is low. That’s exactly what we need.
It should be
First, look at the random model (rand_m). It has roughly the same estimated elasticity in each of its partitions. We can already see just by looking at the plot that it won’t help us much with personalisation since it can’t distinguish between the high and low price elasticity days. Next, consider the predictive model, pred_m. That model is actually promising! It manages to construct groups where the elasticity is high and others where the elasticity is low. That’s exactly what we need.
Model 1 is _elast_mpred; Model 2 is _pred_mpred. Using _mpred is confusing as it can refer either of these.
There is an issue on chapter Evaluating Causal Models, in the following paragraph
First, look at the random model (rand_m). It has roughly the same estimated elasticity in each of its partitions. We can already see just by looking at the plot that it won’t help us much with personalisation since it can’t distinguish between the high and low price elasticity days. Next, consider the predictive model, m_pred. That model is actually promising! It manages to construct groups where the elasticity is high and others where the elasticity is low. That’s exactly what we need.
It should be First, look at the random model (rand_m). It has roughly the same estimated elasticity in each of its partitions. We can already see just by looking at the plot that it won’t help us much with personalisation since it can’t distinguish between the high and low price elasticity days. Next, consider the predictive model, pred_m. That model is actually promising! It manages to construct groups where the elasticity is high and others where the elasticity is low. That’s exactly what we need.
Model 1 is _elast_mpred; Model 2 is _pred_mpred. Using _mpred is confusing as it can refer either of these.