polkas / cat2cat

Handling an Inconsistently Coded Categorical Variable in a Longitudinal Dataset

💫 About Me:

Developer, Statistician, Data Scientist and Agile Team Player.

🌐 Socials:

💻 Tech Stack:

🕸️ Tech Stack:

---

cat2cat's Issues

Check knn

For different type of features

Rewrite stop if not

Provide more informative messages

Cross the probabilities for all methods

cross_p() to assess an average for all methods or to use a certain combination of them.

Ml models performance transparency

Guide how to confirm ml performance. Possibly an optional options like verbose for each ml model

Research parsnip to mimic the python ml solution

python solution using sklearn and inheritance of classes.
https://github.com/Polkas/py-cat2cat/blob/main/src/cat2cat/cat2cat.py
https://github.com/Polkas/py-cat2cat/blob/4b5f0c1099732cba9a65fc14ec03617fbfbeacb0/src/cat2cat/dataclass.py#L127

Write proper uml

nomlnoml

Why we need it -examples

Regression add code variable

Plot

Statistical tests

Support for data.table

Ggplot2 function

A plot to visualize a structure of a groups across time and others

many ml models at once

then not wei_ml_c2c more wei_knn_c2c and wei_rf_c2c. Sb might want to cross this models.
Rewrite the code

Internal validation functions for cat2cat arguments

Transfer cat2cat stopifnot calls to separate validation functions

prune_df() a set of options to reduce replications

Remove zero probability of a row or leave a biggest one or ...

Organize old materials

• prepare case study based on a journal paper materials
• process code from existing ones for unbalanced panel
• example and code - 2 different datasets with the same key
• ...

Boost roxygen2

...

more than 2 periods for each case

examples and vignettes

research icd9 vs icd-10

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3865615/

Plots section README and Vigniette

Wrong df for regression

Nrow(occup_2$old$wei) does not skipping zero probability rows when weighs are used inside lm. So sum of bigger than zero will solve it.

Vigniette

Write a vigniette

Unbalanced and balanced

Change to dataset with/without unique informative identifier.

Cat2cat_agg additional stop of not

Check if only one count for each level

Cat2cat optional freq_var

Here this optional variable will be a vector which contains a frequencies for each group. Why we need this e.g. if sb want to make a recursive weights for frequencies.

Check length , unique values,

cat2cat potenatial siblings

in terms of a jurnal papers and packages

Check if trans is compatible with data

Any levels in data

README graphic

clear and informative. Add some image.

cat2cat frequencies for already replicated data

So the frequencies for already replicated data in cat2cat will assume naive probabilities (equally spaced).
We have to enable usage of weis? ... might sum od weis