Feature/implementation of mimca - #163
Conversation
| from tqdm import tqdm | ||
|
|
||
|
|
||
| def moy_p(V, weights): |
There was a problem hiding this comment.
do not hesitate to give more explicit name for functions
| missing_indices = rng.choice(total_values, n_missing, replace=False) | ||
| row_indices = missing_indices // n_cols | ||
| col_indices = missing_indices % n_cols | ||
| for i in range(n_missing): |
There was a problem hiding this comment.
use a vectorization to avoid loop/
Suggestion :
...
missing_indices = rng.choice(total_values, n_missing, replace=False)
row_indices, col_indices = np.unravel_index(missing_indices, (n_rows, n_cols))
data.values[row_indices, col_indices] = np.nan
return data
| return df_reconstructed | ||
|
|
||
| def imputeMCA( | ||
| don, |
There was a problem hiding this comment.
rename "don" ? Not very an explicit name...
| dict | ||
| Dictionary containing: | ||
| - "tab_disj": Disjunctive coded table after imputation. | ||
| - "completeObs": Complete dataset with missing values imputed. |
There was a problem hiding this comment.
rename completeObs ? usually, variable names in python do not contain capital letter, but _ if multiple "words"
| - "completeObs": Complete dataset with missing values imputed. | ||
|
|
||
| """ | ||
| don = pd.DataFrame(don) |
There was a problem hiding this comment.
why pd.DataFrame(don) if don is already a DataFrame (specified as such in docstring) ?
| if ( | ||
| not pd.api.types.is_numeric_dtype(don[col]) | ||
| or don[col].dtype == "bool" | ||
| ): # noqa: E501 |
| - "completeObs": Complete dataset with missing values imputed. | ||
|
|
||
| """ | ||
| # Ensure the data is a DataFrame |
There was a problem hiding this comment.
suggestion: since it is expected a DataFrame in input, check the type of "don". If not dataframe, report in a log and convert it.
| Z = Z.subtract(Z_mean, axis=1) | ||
| Zscale = Z.multiply(np.sqrt(M), axis=1) | ||
|
|
||
| print("Centered and scaled data (Zscale):") |
There was a problem hiding this comment.
not fan of print.
Use logging instead
| @@ -0,0 +1,665 @@ | |||
| import numpy as np | |||
There was a problem hiding this comment.
add general documentation for the file.
|
|
||
| """ | ||
| if verbose: | ||
| print(f"{print_msg}...", end="", flush=True) |
There was a problem hiding this comment.
logging instead of print
| else: | ||
| row_w = np.array(row_w, dtype=float) | ||
| row_w /= row_w.sum() | ||
| ncp = int(min(ncp, X.shape[0] - 1, X.shape[1])) |
There was a problem hiding this comment.
why shape[0] - 1 ?
I think your tests do not passed because of that.
…eigenvalue shrinking logic
3 participants
No description provided.