| Pitfall | What goes wrong | Fix |
|---|---|---|
scaler.fit(X) on all data before the split | test statistics leak into training; optimistic CV | fit inside a Pipeline; cross-validate the Pipeline |
Scaling before train_test_split in a notebook | same leak, very common | split first, or use a Pipeline |
Separate scalers for train and test (fit_transform on both) | test is scaled by its own statistics: features shift | fit on train, transform test |
| Time series scaled with future data | look-ahead leakage | fit on the training window; refit at retrain |
MinMaxScaler at serving time sees values outside the training range | outputs < 0 or > 1; NN saturation | clip=True, or robust/standard scaling |
| Quantile transform with few rows | jagged mapping; n_quantiles > n warning | n_quantiles=min(1000, n_rows) |
| Box-Cox on zeros/negatives | ValueError | Yeo-Johnson or log1p |
| Scaling binary / one-hot columns for trees | harmless but pointless | passthrough |
| Interpreting standardised coefficients as raw effects | wrong units in reports | report per-SD effects or back-transform |
| Re-fitting the scaler on each prediction batch | inconsistent features between batches | persist the fitted scaler with the model |
| Different preprocessing in training and serving code | train/serve skew | one serialised Pipeline used in both |
| Scaling the target for a tree model and forgetting to invert | predictions in the wrong units | TransformedTargetRegressor |
| Scaling IDs or codes | meaningless distances | drop IDs; encode codes as categories |
| Sentinel values (−999) scaled as numbers | distorts mean/std | convert to NaN + indicator first |
Leakage size: when it matters
Standard scaling leaks only the mean and variance of the test fold: the effect on CV scores is usually small for large datasets. It becomes material when:
- datasets are small (hundreds of rows);
- the transform is flexible (quantile, power with fitted λ, target-dependent steps);
- the test distribution differs from training (time drift), so the leak smuggles in shift information.
Because the fix (a Pipeline) costs nothing, always use it.
Serving checklist
- The fitted Pipeline (scalers included) is serialised and versioned with the model.
- Input columns and order are validated at serving time (
feature_names_in_). - Out-of-range behaviour is defined (clip, or robust scaling).
- Drift monitoring compares serving feature distributions with the training statistics stored in the
scaler (
mean_,scale_,quantiles_).
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.