Comments (2)
_generate_starting_values
and _get_lower_and_upper_bounds
differ in calculation by design. The optimisation itself works on "media unit" but the constraint within it needs to use the budget to make sure the constraint of "using all the budget" is met along the way. That is why the bounds are based on an average of historic media unit and the starting values include the budget in that calculation (also note that starting values dont really need to include budget and we only do that so we can provide the user an average value they can compare against after the optimisation and for it to be a fair comparison they should both use the same budget, although starting values could be random technically).
I think the current behaviour that you are finding is due to #82 (which should be fixed this week). The best thing is to ignore the third output on optimize_media. find_optimal_budgets
and calculate yourself some average budget you want to compare against (and then use predict
to get the prediction for that given media allocation).
Since the prices and bounds are both per channel calculating the bounds in monetary value and then converting back to media unit, or calculating them in media unit directly should yield the same result.
This behaviour does require the user to pass a budget and bounds (percentages) that are aligned. What other interface would you prefer?
Let me know if Im somehow miss-understanding the issue you are finding.
from lightweight_mmm.
Thanks for the swift reply @pabloduque0 actually I will close this comment as my issue was resolved by updating to the latest version of lightweight mmm because in the previous version I had (0.1.5), the bounds were calculated differently.
from lightweight_mmm.
Related Issues (20)
- How can I run kind of gridsearch to find the best custom priors for a hill_adstock or carryover model ?
- How can I input future media_data_test for optimization in upcoming periods? HOT 1
- How can channel-wise optimized conversions be obtained?
- Extra features
- Addressing Heteroscedasticity
- Geo level attribution and response curves
- Budget Allocation Percentage breakdown by channel HOT 10
- Question on tensorflow requirement
- Dtype object is not a valid JAX array type. Only arrays of numeric types are supported by JAX. HOT 5
- Same pre-optimization and post-optimization channel budget allocation ratios , but suggesting much higher budget instead of aligning the budget to the one i requested. HOT 10
- Paid Search bias - Nested model
- Divergences and n_eff
- Outliers and influential points
- budget allocator: How to set up lower bound and upper bound per channel?
- add got incompatible shapes for broadcasting: (95,), (90,). HOT 5
- TypeError: add got incompatible shapes for broadcasting: (58,), (54,). HOT 8
- Rendering of several plots not working! HOT 1
- Negative Values in Pre optimization predicted Target vs Post optimization predicted Target
- Budget Optimization
- RuntimeError: Cannot find valid initial parameters. Please check your model again.
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from lightweight_mmm.