Plot Waterfall Components Decomposition #625
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
Missing cell. Co-Authored-By: Carlos Trujillo <[email protected]>
pymc_marketing/mmm/base.py
Outdated
| dataframe["seasonal"] = ( | ||
| dataframe["sin_order_1"] | ||
| + dataframe["sin_order_2"] | ||
| + dataframe["cos_order_1"] | ||
| + dataframe["cos_order_2"] | ||
| ) | ||
| dataframe.drop( | ||
| ["sin_order_1", "sin_order_2", "cos_order_1", "cos_order_2"], | ||
| axis=1, | ||
| inplace=True, | ||
| ) |
There was a problem hiding this comment.
We should have a function that extract the seasonal components by looping over "cos_order_" and "sin_order_" over all components defined by yearnly_seasonality (https://github.com/pymc-labs/pymc-marketing/blob/main/pymc_marketing/mmm/delayed_saturated_mmm.py#L54)
There was a problem hiding this comment.
Yes, sounds good. Should we created in this same PR?
There was a problem hiding this comment.
I did the following and looks enough:
if getattr(self, "yearly_seasonality", None):
contributions_fourier_over_time = pd.DataFrame(
az.extract(
self.fit_result,
var_names=["fourier_contributions"],
combined=True,
)
.mean("sample")
.to_dataframe()
.squeeze()
.unstack()
.sum(axis=1),
columns=["yearly_seasonality"],
)There was a problem hiding this comment.
yeah, you can do it on this PR
Co-Authored-By: Carlos Trujillo <[email protected]>
Co-Authored-By: Carlos Trujillo <[email protected]>
Co-Authored-By: Carlos Trujillo <[email protected]>
|
I like this plot a lot! Is it possible to flip it around so that the most impactful component is shown first on the X-axis? In visualization theory, our eyes start from the top-left and scan to the right, much like reading a book. If doable, it's best to orient by importance in the same manner. |
|
@cetagostini-wise I have seen you have a conflict with the mmm example notebook. Git with notebooks is a nightmare so here is a tip: If you want to rebase and keep the changes in your branch then you run: If you want to keep the changes from I hope it helps :) |
pymc_marketing/mmm/base.py
Outdated
| @@ -1274,6 +1276,112 @@ def plot_channel_contribution_share_hdi( | |||
| def graphviz(self, **kwargs): | |||
| return pm.model_to_graphviz(self.model, **kwargs) | |||
|
|
|||
| def plot_waterfall_components_decomposition( | |||
| self, original_scale: bool = True, figsize: Tuple = (14, 7), **kwargs | |||
There was a problem hiding this comment.
can we pass an optional ax like in the other plots
Co-Authored-By: Carlos Trujillo <[email protected]>
Co-Authored-By: Carlos Trujillo <[email protected]>
As discussed can we do it in another branch? |
Description
Related Issue
Checklist
Modules affected
Type of change
📚 Documentation preview 📚: https://pymc-marketing--625.org.readthedocs.build/en/625/