00:12:14 Daniel Chen: here isn't an echo on my end. not relaly
00:26:36 Daniel Chen: so far we've created a simulated set of values with random noise right?
00:26:52 Ildiko Czeller: yes
00:27:13 Ildiko Czeller: for a classification problem
00:27:13 Daniel Chen: and now we're fitting a Bayesian model on data? am I following that correctly?
00:27:33 Daniel Chen: so the packages mentioned in the beginning is for Bayesian stuff?
00:27:53 Ildiko Czeller: yes, i think stan does bayesian prediction.
00:28:24 Ildiko Czeller: probably will compute tthe equivocal zones later if you mean that by the package mentioned in the beginning
00:28:32 Daniel Chen: i mean this chapter as a whole is using Bayesian models to see how much we should "trust" predictions
00:28:52 Daniel Chen: yeah. ok. so this is all Bayesian specific stuff?
yes stan is for Bayesian stuff
00:29:18 Ildiko Czeller: equivocal zones can be calculated for non bayesian models as well I think
00:29:45 Daniel Chen: what's the data_grid? i think i just missed it
00:30:58 Ildiko Czeller: i think it is just your simulated dataset with x, y as predictors, isn't it?
00:31:29 Daniel Chen: oh it looks like the predicted values? classes and probablilties. similar to inputs used for yardstick
00:31:33 Ildiko Czeller: the equivocal zones they are not so sophisticated for bayesian models in my understanding
00:31:44 Daniel Chen: oh no. yeah it looks like simulated data
00:32:16 Ildiko Czeller: I meant for NON bayesian models they are less sophisticated
01:02:56 Daniel Chen: can you go back to what the pca_stat value is when being compared to the Chicago data?
01:11:23 Daniel Chen: what's the pca stat values?
01:11:36 Daniel Chen: why is it 1 column of values when you have multiple PCs?
01:13:00 Daniel Chen: this is in the score function call.
01:13:43 Daniel Chen: oh it's distance from center
01:15:00 Daniel Chen: oooh it's all 9
01:15:14 Daniel Chen: ok that part makes sense that's regular PCA results
01:15:45 Daniel Chen: yeah that's good.
01:16:39 Daniel Chen: i might be moving next 2 weeks.
01:16:45 Daniel Chen: so i might not be availviale
01:17:11 Federica Gazzelloni: thanks
19.3.3 Cohort 4
Meeting chat log
00:26:35 Federica Gazzelloni: The reportable rate is calculated as (n_not_equivocal / n).