DMP Statistical Solutions

Statistical consulting

Statistical Modelling

Our consultants are highly qualified statisticians, expert in a range of analysis areas. A representative range of areas are:

  • Basic statistical models and testing e.g. Generalized Linear Models (GLMs), simple non-parametric methods.
  • More advanced variants: Generalized Additive Models (GAMs), non-linear regression.
  • Sampling design and optimisation.
  • General multivariate probability models: Copulas and the Iman-Conover method.
  • Surface smoothing: multidimensional splines (e.g. TPS, penalised TPRS, geodesic smoothing), kriging.
  • Solutions for difficult data: imputation, ridge and principal components regression.
  • Repeated measures: Generalized Linear/Non-linear Mixed Models, Generalized Estimating Equations (GEEs).
  • Multivariate methods e.g. dimension reduction (PCA, MDS etc), multivariate tests.
  • General optimisation: supply chain, delivery routes, linear/non-linear optimisation.
  • General model selection: optimising model complexity – all possible subsets, standard step methods, AIC/QIC/BIC, k-fold CV, lasso, etc.
  • Datamining: analysis of large datasets, predictive models for classification (e.g. fraudulent transactions, churn, network faults), tree-methods, boosting (incl. gradient boosting), market-basket, neural networks etc.
  • Computer intensive methods: bootstrapping, cross-validation, randomisation/permutation testing.
  • Bayesian inference.

case_study_1
case study

Customer / Brand Equity Modelling

DMP provide annually updated statistical models for this software package, based on annually revised survey information

read more…
case_study_2
case study

Environmental
Impact

For example, the Strangford Lough data comprise of continual observer sightings of marine mammals and birds

read more…
case_study_3
case study

Cosmeceutical
Trial Analysis

DMP provide the analysis for regular cosmeceutical trials conducted by a UK-based trials unit, the end clients

read more…