Demand forecasting or cost estimation model by Göksu DemirciDemand forecasting or cost estimation model by Göksu Demirci
Demand forecasting or cost estimation modelGöksu Demirci
Cover image for Demand forecasting or cost estimation model
You have operational history in a database or a pile of spreadsheets, and decisions are still being made by feel. I turn that history into a model that gives a number, and into a service your systems can call.
What you get: a data audit first, where I tell you honestly whether your data can support the question you are asking, before you pay for the rest. Feature engineering and model selection, with the evaluation metric chosen for your cost of being wrong, not for a leaderboard. Honest performance reporting, including where the model fails. Deployment as a FastAPI service in Docker, plus the notebook and the code.
Where I have done this: demand and cost estimation models running daily in a manufacturing ERP I built and operate. A production-line quality classifier on 1.18M rows with a 0.44% positive rate, where naive accuracy is a trap and recall is the only number that matters.
If the data audit shows the problem is not solvable with what you have, I say so in week one and you pay for that week only.
Starting at$1,200
Duration3 weeks
Tags
Data Engineer
Service provided by
Göksu Demirci Konya, Turkey
Demand forecasting or cost estimation modelGöksu Demirci
Starting at$1,200
Duration3 weeks
Tags
Data Engineer
Cover image for Demand forecasting or cost estimation model
You have operational history in a database or a pile of spreadsheets, and decisions are still being made by feel. I turn that history into a model that gives a number, and into a service your systems can call.
What you get: a data audit first, where I tell you honestly whether your data can support the question you are asking, before you pay for the rest. Feature engineering and model selection, with the evaluation metric chosen for your cost of being wrong, not for a leaderboard. Honest performance reporting, including where the model fails. Deployment as a FastAPI service in Docker, plus the notebook and the code.
Where I have done this: demand and cost estimation models running daily in a manufacturing ERP I built and operate. A production-line quality classifier on 1.18M rows with a 0.44% positive rate, where naive accuracy is a trap and recall is the only number that matters.
If the data audit shows the problem is not solvable with what you have, I say so in week one and you pay for that week only.
$1,200