Articles, newest first
Clustering Is a Model of Similarity, Not a Discovery of Ground Truth
A clustering algorithm always answers a question, but the question is partly specified by us. Changing scale, distance, representation or objective can chang...
Read article
Aspartame, Fruit and the Difference Between a Relevant Fact and a Complete Safety Argument
The metabolites of aspartame are chemically ordinary, and that matters. It is not by itself a proof of safety, because fruit also contains things that are to...
Read article
Forecast Value Added: Is Your Process Helping?
Forecasting processes accumulate steps: a statistical model, a planner override, a consensus meeting. Each is assumed to improve the number. FVA is how you f...
Read article
Temporal Hierarchies: Reconciling Across Time Granularities
A hierarchy does not have to be geographic. Aggregating a series over time produces the same coherence problem, and the same machinery solves it.
Read article
LLM Quantization Is Not Just Using Fewer Bits
Saying that a model is 4-bit tells you how weights are represented, not how they were calibrated, which tensors were quantized, what arithmetic the hardware ...
Read article
Label Noise in Supervised Learning: When the Target Cannot Be Trusted
Label noise is one of the most damaging data quality problems in supervised learning because it corrupts the target the model is trained to imitate.
Read article
Neural Forecasting: What the Architectures Actually Do
Neural forecasting has produced genuinely useful architectures and a great deal of noise. The differences between them are more interesting than their benchm...
Read article
Modelling Count Time Series
Daily incident counts are integers, non-negative, often small, and correlated with yesterday. ARIMA assumes none of that and Poisson regression assumes indep...
Read article
Long Memory and Fractional Integration in Time Series
Standard practice offers two options: the series is stationary, or you difference it. Some series are genuinely in between, and forcing them either way loses...
Read article
Dynamic Time Warping and Time Series Clustering
Two series can trace an identical shape while one runs slightly ahead of the other. Point-by-point distance calls them dissimilar; dynamic time warping does ...
Read article