Can you build a custom prediction model for my organisation?
Yes. We build models on your own historical data and work with clients across Greece and the EU, in Crete or remotely.
How do you explain what the model does?
Every model comes with an explanation of its decisions: which factors matter most overall, and why each individual prediction came out the way it did. You can show this to management, customers or auditors.
How is the model delivered?
As a script that generates predictions with the model from new data in the same format as the training data. You run it on your own computers or servers, so the model always stays in your systems. If you want it built into your own software, we can integrate it into your code at extra cost, agreed in the written scope.
How is the scope agreed?
Always in writing, in three steps. You send a description, your desired timeline and whether a budget is available. We reply with questions and a written scope and timeline, which we adjust together. Once both are settled, we send a fixed quote. The contract contains the agreed scope and the fixed quote and takes effect when it is signed and the first payment is made. Work starts once we receive the data.
What if we need changes later?
It depends on their size. Small changes (up to about 4 hours of work) are charged at €55/hour + VAT, with a written estimate before we start. Larger changes, such as new features, new data or anything that changes the scope or timeline, go through the standard process as a new request: written scope, fixed quote and a contract addendum. The original scope and price stay protected.
How much data do I need?
It depends on what you want to predict. A few hundred well-recorded historical cases can be enough for a first model. During the written scoping we review a description or sample of your data and tell you whether it is sufficient before you commit.
How do I know the model's accuracy is real?
Every model comes with accuracy estimates and confidence intervals from repeated cross-validation, corrected for the optimism that comes from picking the best of many candidate models. You see the realistic figure, not the best-case one.
What methodology do you follow?
Data projects follow CRISP-DM: business understanding, data understanding and preparation, modelling, evaluation and deployment. Standard packages run as a fixed-price process. Custom projects add milestones and feedback rounds.
Do you work with health or omics data?
Yes, under a data-processing agreement and with confirmation of the ethics approval or legal basis for the data. Every analysis comes with a report that documents each step, so it can be reproduced.
Does our data leave our company?
It does not have to. The finished model and any data pipelines always run on your own infrastructure. Where we process data ourselves, it is covered by an NDA and a data-processing agreement.
What does it cost?
Packages start from €1,800 + VAT for a model audit, €4,500 + VAT for predictive analytics and €6,000 + VAT for feature discovery. The price depends mainly on the size and complexity of the data available. Custom data pipelines are priced in the written scope. The final price is fixed in a written quote once the scope and timeline are agreed.
