Daniele Baranzini

• PhD
• Head at HAIKAI

The Before After Machine (BAM) is a meta-learning algorithm where a classificator is required to operate a prediction, in the future expected temporal window assigned by a temporal forward forecast. "It returns a prediction from the future" Daniele Baranzini, PhD

The TWISTER (FDM)* algorithm is generated to predict a flight risk (target) by capturing the temporal dimension of flight operations data in a flight shapelet, a flight profile on one or more features over time. This profile information might otherwise be lost when testing predictive models capacity on tabular data disregarding the time varying d...

Generative AI and Large Language Models 2023

BowTie Predictive is contained within a web app called BowtieRegister, which can be found at www.bowtieregister.com. “Introducing our new app that harnesses the power of AI at the core of a bowtie structure to deliver unparalleled performance and insights in risk assessments. The bowtie is quantified on real data and not generic estimates, with t...

The assessment of cognitive deficits is pivotal for diagnosis and management in patients with parkinsonisms. Low levels of correspondence are observed between evaluations assessed with screening cognitive tests in comparison with those assessed with in-depth neuropsychological batteries. A new tool, we named CoMDA (Cognition in Movement Disorders A...

There is a strong and growing interest in using the large amount of high‐quality operational data available within an airline. One reason for this is the push by regulators to use data to demonstrate safety performance by monitoring the outputs of Safety Performance Indicators relative to targeted goals. However, the current exceedance‐based approa...

Airline applications of AI methods in real settings: optimize fuel consumption and safety events. Obtaining a Fuel Precision Approach by deployment of Statistical and Machine Learning algorithms.

The Pattern AGgregating Ensemble algorithm (PAGE) maps a validated* machine-learning model into a simpler aggregated structure where multiple predictions are reduced to individual expectations. Both target and predictors are reduced into smaller indicator classes. A statistical method then tests if such new aggregated model, the “paged” model, pres...

TWIST algorithn for aviation flight operations. 1) Determining when to make safety predictions, not only what to predict in flight operations. 2) Maximising the use of FDM data in flight operation performances

Mindful organising is a key integrating concept in resolving the organisational accident. Mindful organising is both the unique source of critical information about the normal operation, as well as the key recipient of intelligence about the operation, ensuring that operational actions are always informed by the most current, relevant information a...

Backward piping algorithm: loss (or increment) in permissible operations for any data-frame parsed between two different software applications Dr Daniele Baranzini (Ergonomica) [ video explanation given @ https://youtu.be/Rk3abMXds2Q ]

Beyond machine learning prediction for Aviation Big Data. Model Deployment devices and exemplary model exploitation for fuel consumption in flight operations.

Introduction: Subjective cognitive decline (SCD) in older adults are an early risk indicator for Alzheimer's disease or other forms of dementia, making older adults with SCD a target population for proactive interventions. The aim of this study was to determine if perceptual-cognitive training (PCT) can serve as a proactive intervention and enhance...

SPSS Single dataframe aggregating SPSS Multiply Imputed split files

Machine Learning applications in Commercial Aviation

In my previous TechNotes (Baranzini, 2018) I described some Machine Learning applications over tens of thousands real flight operations data out of a GDPR protected airline in 2017. But you do not always need data to inform your decision-making in aviation operations, although your model may still be based on numbers.

LINK 50 days free for Paper here: https://authors.elsevier.com/a/1XVLj7t2zYxTg4 ABSTRACT The purpose of this study is to review and present the contribution of the Joint Research Centre of the European Commission to the development of a new methodological approach for measuring the effective capacity of national programmes aimed at reducing the ri...

This TNote describes how to replace Bow Ties structures by fitting Machine Learning Algorithms in order to generate new (or confirm!) Bow Tie structures based on full big data evidence instead of naïf expert judgment risk estimates. More efficient and effective "in-split" * formalizations (e.g. replacing Boolean logic for splitting Bow-Tie branches...

Algorithmic results of Machine Learning applications in a commercial aviation airline. Scores for Unstable Approaches (UA) for individual flight operations of an undisclosed european airline are presented. A selection of features (predictors) have been modelled by logistic functions to target the UA variable (prediction). All possible combination...

A "risk pattern" informally is a configuration of variable values coinciding with an event of interest (Baranzini, 2015). Formally a risk pattern denotes a "generic predictor machine algorithm".

real Big Data applications in aviation

Perspectives and best practices on system-wide organisational changes. Implications for Research and International Regulatory bodies in Chemical Accident Prevention and Preparedness. Relevance is given to the ”time span” affecting system wide large changes and methods to minimize uncontrolled change growth rates and speeds.

We examine the use of role-switching as an intrinsic motivational mechanism to increase engagement in long-term child–robot interaction. The present study describes a learning framework where children between 9 and 11-years-old interact with a robot to improve their knowledge and habits with regards to healthy life-styles. Experiments were carried...

This report is a limited distribution publication because of information provided in confidence to the JRC by EU Neighbourhodd countries and the sensitivity of national capacity rankings for chemical accident prevention and preparedness assigned to each country based on that information. It has been produced by JRC G05 for DG ECHO A5 under the Admi...

The paper presents new system risk prediction concepts for the aviation system. The concepts of “risk pattern” and “risk intelligence” are a new technology combining Data Science with Data Integration across Airlines, Airports and ATM systems. The role of risk patterns for commercial aviation systems is explored and results are discussed with case...

The EU Vision 2020 sets a goal of reducing the air travel accident rate by 80%. Achieving this vision requires innovation and a different approach. PROSPERO (Proactive Safety Performance for Operations) is an EU FP7 project that will provide an advanced systemic methodology for managing the improvement process to help achieve that goal, as well as...

A study was performed on the status of Natech risk reduction in Europeon Union Member States by means of a questionnaire survey. The results show that natural hazards are increasingly recognised as a possibly important external risk source for chemical facilities. The management of Natech risk is mainly addressed through the Member States’ legal fr...

Lombardy is one of the most densely populated and industrialized regions in Europe, where nearly 280 Seveso sites are located. The issue of risk communication, as set by the European Seveso Directive is therefore of high relevance in this region. Nevertheless, the Lombardy Region Authorities consider that the implementation of the Directive’s provi...

This paper reports on human factors data traceability and analysis of the European Community’s Major Accident Reporting System (MARS). This is the main EU instrument to major accident data collection, analysis and dissemination for process industry according to the provisions of the Seveso II Directive. To date, the MARS database counts approximate...

The importance of teamwork in aviation maintenance has increasingly been recognised over the past two decades in recognition of the demands of managing and organising more and more complex work processes and even more difficult communication and interaction patterns within and across distributed work units, departments and functions. This paper foc...

Real Commercial Aviation environments require big data and machine learning models with high accuracy rates. This is first for safety predictions and then for commercial implications. Ideally any predictive algorithmic models devised by ML or Statistical Learning would require always training and test set error rates measurements. Costs of mis-clas...