Why they matter for selection
- L1/lasso picks one member of a correlated group, often arbitrarily, and the choice flips between folds.
- Trees split on whichever member is marginally better in each bootstrap/fold; importance is spread.
- Permutation importance under-states each member: shuffling one leaves its twin to carry the signal.
- Prediction quality is usually unaffected; interpretation and stability are.
Finding groups
import numpy as np
import pandas as pd
from scipy.cluster import hierarchy
from scipy.spatial.distance import squareform
corr = X_train.select_dtypes("number").corr(method="spearman").abs()
dist = 1 - corr
link = hierarchy.ward(squareform(dist.values, checks=False))
cluster_ids = hierarchy.fcluster(link, t=0.3, criterion="distance") # tune t; ~ |rho| > 0.7 groups
groups = pd.Series(cluster_ids, index=corr.columns).groupby(level=0).apply(list)filter_select.py --corr 0.9 does single-linkage grouping at a fixed |Spearman| threshold and keeps the member
with the strongest univariate signal.
Choosing a representative
| Criterion | When |
|---|---|
| Strongest validated signal | prediction focus |
| Cheapest to compute / most available at serving | production focus |
| Most interpretable to stakeholders | explanation focus |
| A composite (sum, ratio, first principal component of the group) | keeps information from all members |
Grouped permutation importance
import numpy as np
from sklearn.metrics import average_precision_score
def grouped_permutation_importance(model, X, y, groups, n_repeats=10, seed=0):
rng = np.random.default_rng(seed)
base = average_precision_score(y, model.predict_proba(X)[:, 1])
out = {}
for name, cols in groups.items():
drops = []
for _ in range(n_repeats):
Xp = X.copy()
idx = rng.permutation(len(X))
Xp[cols] = X[cols].values[idx] # shuffle the whole group together, keeping its internal correlation
drops.append(base - average_precision_score(y, model.predict_proba(Xp)[:, 1]))
out[name] = (np.mean(drops), np.std(drops))
return outVariance inflation factor (linear models)
VIF_j = 1 / (1 − R²_j), where R²_j regresses feature j on the others. VIF > 5-10 signals multicollinearity
that destabilises linear coefficients (not predictions). statsmodels.stats.outliers_influence.variance_inflation_factor.
Practical policy
- Remove exact and near-exact duplicates (|rho| > 0.98) always.
- For interpretation-heavy models (scorecards, GLMs), keep one per group at |rho| > 0.8.
- For GBMs used for prediction, keep groups unless serving cost matters; report group-level importance.
anomaly-detection-model-builder
Build and evaluate anomaly detection for tabular, sensor and transaction data with few or no labels: define the anomaly type (point, contextual, collective), choose detectors (robust z-score/MAD and IQR rules, Mahalanobis with robust covariance, k-NN distance, LOF, Isolation Forest, one-class SVM, PCA reconstruction / autoencoders, forecasting residuals for time series), combine them by rank, explain each alert by its deviating features, set thresholds from an alert budget or the score distribution, and evaluate with precision@k, PR-AUC and recall on known incidents, or detector agreement when no labels exist. Includes pure-Python detectors (robust z, Mahalanobis, kNN, Isolation Forest, ensemble) and an evaluation script.
catboost-model-builder
Build CatBoost models that use its strengths correctly: raw categorical columns as cat_features (ordered target statistics, no manual encoding), text_features, Pool construction with NaN-safe categories, ordered boosting and has_time for temporal data, symmetric (oblivious) trees, overfitting detector with eval_set and use_best_model, losses for classification, regression (Quantile, Poisson, Tweedie) and ranking (YetiRank), class weights, monotone constraints, feature importance (PredictionValuesChange, LossFunctionChange) and SHAP values, GPU training and model export. Plans cat/text/numeric columns from a CSV and generates a runnable training script, and checks parameter sets for the conflicts CatBoost rejects (iterations vs n_estimators, Bayesian bootstrap with subsample, multiple class-weight options, depth > 16, use_best_model without eval_set).
categorical-encoding-selector
Choose and implement the right encoding for every categorical column given the model family and the data: one-hot (with infrequent-level grouping and handle_unknown), ordinal with an explicit order, cross-fitted target encoding (scikit-learn TargetEncoder, CatBoost ordered statistics, category_encoders inside CV), frequency/count encoding, feature hashing, native categorical support in LightGBM, HistGradientBoosting, XGBoost and CatBoost, prefix levels for codes (zip3, NAICS-2), and embeddings. Measures cardinality, rare-level share, unseen-level rate on a time-ordered holdout and target signal per column, emits the ColumnTransformer, and quantifies target-encoding leakage (in-sample vs out-of-fold vs holdout AUC).
classification-model-selector
Choose the right classification algorithm for tabular, text or mixed data and prove the choice with a fair bake-off: logistic regression (L2, L1, elastic net), linear SVM, naive Bayes, k-NN, RBF SVM, decision trees, random forest, extra trees, HistGradientBoosting, LightGBM, XGBoost, CatBoost, explainable boosting machines (EBM/GAM), points scorecards and MLPs, matched to data size, width, categorical and high-cardinality share, missingness, class imbalance, text, and constraints such as interpretability, monotonic constraints, calibrated probabilities, latency budgets and tiny deployable models. Ranks candidates with starting hyperparameters and cautions, then generates a bake-off script running every candidate on identical CV folds with per-fold metrics, fit time and prediction latency.
clustering-algorithm-selector
Choose the right clustering algorithm and check that clustering makes sense at all: cluster tendency (Hopkins statistic with a correlated-features caveat, nearest-neighbour isolation), then k-means / MiniBatchKMeans, Gaussian mixtures with BIC, agglomerative (Ward, average, complete, single) with dendrograms, DBSCAN, HDBSCAN, OPTICS, spectral clustering, BIRCH, mean shift, k-prototypes and k-modes for mixed or categorical data, and Gower distance, matched to data size, expected shapes, noise, varying density, known k, soft memberships, hierarchy and streaming. Covers preprocessing (scaling, encoding, dimensionality, distance metrics), validation (internal indices, bootstrap stability, external labels, business profiling) and assigning new points.
collaborative-filtering-implementer
Implement collaborative filtering recommenders that beat popularity: item-based kNN with cosine, Jaccard or asymmetric similarity, shrinkage, neighbourhood size, positive-only or mean-centred rating signals and 'because you liked X' explanations; implicit ALS with confidence weighting, BPR pairwise ranking and the closed-form EASE model in plain NumPy, tuned on a temporal validation split by recall@k and NDCG@k and refit on all data; plus user-kNN, explicit matrix factorisation, the implicit/LightFM libraries, hyperparameter ranges, cold-start fallbacks, scaling with sparse matrices and approximate nearest neighbours. Produces top-k lists in the format of the recommendation evaluator.