casact / rp-bnn-claims Goto Github PK
View Code? Open in Web Editor NEWIndividual Claims Forecasting with Bayesian Mixture Density Networks
License: Mozilla Public License 2.0
Individual Claims Forecasting with Bayesian Mixture Density Networks
License: Mozilla Public License 2.0
I'm an renv
novice, so apologies if this is an obvious question. When running renv::restore
, I get the message * The library is already synchronized with the lockfile.
. Yet RStudio tells me that I don't have recipes
, tfdatasets
and two others. I know that I have recipes
installed globally, but it doesn't look as though it appears in the file renv.lock
.
Does the lockfile simply need to be updated?
@PirateGrunt added you as admin, not sure if that's enough to transfer ownership but worth a shot (so I don't need to obtain unnecessary permissions on the cas org)
Hello,
I'm working on individual claims reserving and your work is incredible ! Thank you so much for sharing your code.
I came across an error and I can't fix it (is the problem coming from the new version of tfprobabilty or tensorflow or something else ?)
It's related to this lines of code (model.R) :
paid_out <- out_sequence %>%
layer_dense_variational(units = 4,
make_posterior_fn = posterior_mean_field,
make_prior_fn = prior,
kl_weight = 1 / n_rows,
activation = "linear") %>%
layer_distribution_lambda(
function(x) {
d <- tfd_mixture(
cat = tfd_categorical(logits = x[,,1:2]),
components = list(
tfd_transformed_distribution(
tfd_log_normal(x[,,3], 1e-3 + ln_scale_bound * k_sigmoid(scale_c * x[,,4])),
tfb_affine_scalar(shift = -1e-3)
),
tfd_deterministic(loc = k_zeros_like(x[,,3]))
)
)
},
name = "paid_out_"
)
Am getting this error :
Warning message in tfb_affine_scalar(shift = -0.001):
“tfb_affine_scalar() is deprecated, please usetfb_shift(shift)(tfb_scale(scale))
instead”
Error in py_call_impl(callable, dots$args, dots$keywords): RuntimeError: Exception encountered when calling layer "paid_out_" (type DistributionLambda).Evaluation error: TypeError: got an unexpected keyword argument 'use_static_graph'
Detailed traceback:
File "/miniconda/envs/r-reticulate/lib/python3.8/site-packages/decorator.py", line 231, in fun
args, kw = fix(args, kw, sig)
File "/miniconda/envs/r-reticulate/lib/python3.8/site-packages/decorator.py", line 203, in fix
ba = sig.bind(*args, **kwargs)
File "/miniconda/envs/r-reticulate/lib/python3.8/inspect.py", line 3037, in bind
return self._bind(args, kwargs)
File "/miniconda/envs/r-reticulate/lib/python3.8/inspect.py", line 3026, in _bind
raise TypeError(Call arguments received:
• inputs=tf.Tensor(shape=(None, 11, 4), dtype=float32)
• args=<class 'inspect._empty'>
• kwargs={'training': 'None'}
Thank you so much !
After addressing the renv
issue #14, I'm now having an issue with piggyback::download()
. It worked earlier, but I'm not getting the error:
Error in df[update, ] : incorrect number of dimensions
In addition: Warning message:
In get_token() : Using default public GITHUB_TOKEN.
Please set your own token
Is the repo missing a .pbattributes file?
In requirements.txt, line 14: pkg-resources==0.0.0
I do not believe this to be a valid PyPI package number.
To the original author(s), please ensure that package numbers are valid in order for users to build the project.
I'll try to create a PR to correct this and any other package version issues soon.
Thank you! :D
I tried to build the repo and received an error, stating that PyPI cannot find the version of opt-einsum
at 3.0.0.
could not find function "layer_distribution_lambda"
file model.R
I use keras 2.13.0
tensorflow 2.14.0.9000
trims
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.