Comments (12)
You will have to input your own Model ID to get the reallocated result for that model ;) when you look at the result one-pagers, you'll see a ID in the title. Input that into the allocator.
from robyn.
Would you mind sharing a bit insight of if one or some the model results and the reallocation make sense? That would be really helpful:)
from robyn.
Ah, regarding Lower and upper bounds for the allocator, it needs to be exact the same order as your media variables defined.
In the sample Code, I put 5 channels. So the lower and upper bounds both have 5 values.
from robyn.
Hi, I double checked and I was using a model ID from my own data's iterations, and the lower/upper bounds were in the exact same order, so I'm a bit unsure what's going wrong? I am happy to share anything with you's btw (though with this test the data is not complete, and I need some better baseline variables...) :-) All help is appreciated, and I am currently in convo with FB's marketing science team too, in Sweden, regarding getting some support with testing this on a client, and with a plan to push towards some case studies (MMM is not prevalent in Norway and VERY expensive... so Robyn, and it's automation, could be a big deal, especially for agencies without employees specialised in the subject)
from robyn.
Could you provide more info on your error? Screenshots of your budget allocator and set_mediaVarName would be great
from robyn.
from robyn.
Hopefully that helps? :-)
from robyn.
Yes, you only have 3 media channels. So you should only need 3 numbers as lower bounds and 3 numbers as upper bounds for the allocator. You're using the default, which has 5 values each. Read the comments behind the function also helps with how to set the bounds
from robyn.
Oh I am blind. Thanks for confirming. So I just removed two values then, and then I got the below:
Error in f.budgetAllocator(modID = "34_79_5", scenario = "max_historical_response", :
provided modID is not within the best results
I was also wondering if the model one pagers (aka, 40 different PNGS for each iteration) have an order of relevance? Aka, most of the r ratings are basically the same (and mape_lift always equals N/A), but the results can vary massively. With the various models, does Robyn know which one it feels fits best?
Thanks again
from robyn.
Please run model_output_collect$allSolutions, or look at the Model IDs in the onepager title, to get correct IDs
from robyn.
regarding "which one it feels fits best?": all printed results are Pareto-optimal solutions with trade-offs between the two optimisation objective functions: NRMSE (model error) and DECOMP.RSSD (business logic). According to Wikipedia, pareto-optimal means "no individual can be better off without making at least one individual worse off". In other words, all printed solutions can't be better trading off the two objectives. On the other hand, we're also returning a set results instead of just "the best one" on purpose, so that users can relate their business reality to different model outcomes and select the most suitable one. We don't make that choice for you.
from robyn.
there's been couple of fixes pushed. please try out the latest version and let us know if everything works.
from robyn.
Related Issues (20)
- Issue downloading Nevergrad HOT 1
- Error in { : task 1 failed - "object 'x_imme' not found"
- Error in Robyn Refresh HOT 4
- Clarify `hyper_names` documentation HOT 4
- Simulation for GRP at campaign and format level HOT 3
- Total response in Budget allocation HOT 7
- Understanding train/test/split HOT 3
- Understanding Input for Response Curves HOT 2
- Segfault caused when running 12k iters and 12 trials HOT 3
- Robyn Allocator Function Prioritizing Minimization Over Optimization HOT 3
- Robyn Allocator: Can't find `stat` called 'align' HOT 3
- Error: unexpected symbol in: date_var
- Error in UseMethod("mutate") & Error in robyn_outputs(InputCollect, OutputModels, pareto_fronts = "auto" HOT 1
- Immedaite vs Carryover Response and bootstrapped ROl Plots HOT 1
- Zero Spend Channels Included in Allocator HOT 6
- Adjusted R2 for test data should not be used HOT 1
- Error in check_hyper_limits(hyperparameters_ordered, "thetas") HOT 3
- Why spend shares % from the one pager do not match with the Initial Avg. Spend Shares % that we get from the budget allocator? HOT 2
- Channels with Coefficients of 0 in the Budget Allocator HOT 2
- Grps instead of the spent HOT 1
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 robyn.