Review Algorithms

In recent years, much has been written about discriminatory, incorrect or unfair data analysis solutions and the use of algorithms. LinkedIn, magazines and newspapers are full of bad examples. Think of the recent Allowances affair by the Tax Authorities and the use of algorithms lead to discrimination by local municipalities. There is social unrest and there are concerns about how data results in decisions.

In recent years, much has been written about discriminatory, incorrect or unfair data analysis solutions and the use of algorithms. LinkedIn, magazines and newspapers are full of bad examples. Think of the recent Allowances affair by the Tax Authorities and the use of algorithms lead to discrimination by local municipalities. There is social unrest and there are concerns about how data results in decisions.

  • We use smart technology in our services,
  • we are committed to fair algorithms’, but also to “well-functioning, effective algorithms” that are used in a business setting,
  • We offer you and your stakeholders confidence,
  • We take your team with us, led by an experienced auditor,
  • We give you the confidence to execute your strategy with modern technology.

Companies and organizations

We also see within companies and organizations that the use of data analysis and algorithms does not always result in achieving the pre-formulated business objectives. This sometimes to great disappointment

One thing is certain, the further digitization of business processes and management decisions is not possible without the application of data analysis and the further use of algorithms. We help you as CFO, Chief Data Officer, controller, DGA (and others) with testing reliable data analysis and algorithm solutions. This allows you to use them with confidence and build trust for your stakeholders; Open Up The Analytics & Algorithms.

Framework

Various initiatives have been launched worldwide to assess algrotimes. In the Netherlands, in Europe, in the US and in other parts of the world. In the Netherlands, the Court of Audit has developed a standard model. A model gives direction to the assessment of an algorithm.

The Court of Audit’s testing framework consists of 5 perspectives on algorithms. The most important risks have been formulated for each perspective. Each risk is associated with a research question. By answering the questions and giving a score, it becomes clear to the user to what extent the risks are limited with the chosen algorithm. How big the risks are for a specific algorithm depends on 2 things:

  • the extent to which advanced techniques are applied in the algorithm,
  • the impact of the algorithm on the entrepreneur, citizen and other stakeholders

We use several standard frameworks within Coney, depending on the setting. Naturally, we coordinate the frameworks in advance for the purpose of assessing the algorithm.

Example Algorithm for determining sales discount for customers

A retail company has developed a new algorithm, essentially a decision tree for granting discounts to online customers. The amount of the discount and the frequency with which discounts are granted partly depend on the spending of customers per month and per product (turnover and margin).

The decision tree included a critical error. The best customers received a relatively lower discount than customers who sporadically made online purchases on the website. The relative margin for the latter group was ultimately negative. We parsed the algorithm and fixed the error.

In 3 clear steps

Our method

1

We define and research questions

We define the research questions in advance and a team of data analysts and data science professionals, led by an experienced auditor, will get to work with your own team. A lot of confidence is added to the correct operation by testing before and during the deployment of a data solution and/or algorithm.

2

Knowledge sharing

We combine our specific knowledge of themes (e.g. risk & control), domains (e.g. finance, audit, risk management) and sectors with knowledge of technology and data analysis. We give you the confidence to execute your strategy with modern technology.

3

Data analysis

As a Coney Minds team, we have been active since 2005 at the intersection of data analysis and audit. In recent years we have been working on integrating process mining, machine learning and data visualization in our services.

Other services