I will audit your data for real predictive signal

Romania

I speak English, Romanian, French

Applied Mathematician

I am an applied mathematician who uses statistics, machine learning, and software engineering to solve real-world problems. My work ranges from determining what can actually be learned from data to de...
About this Gig

Not every dataset contains useful predictive signal, and a more complex model cannot create information that isn't there.

I will investigate your data to determine what can actually be learned, how strong the signal is, and whether it generalizes beyond the data used to discover it.

Depending on the problem, I may use statistical modeling, machine learning, Bayesian methods, regression, survival analysis, time-series methods, simulation, or other appropriate techniques. The method follows the problem, not the other way around.

My work may include:

  • Predictive signal assessment
  • Out-of-sample validation
  • Statistical analysis and diagnostics
  • Model comparison
  • Uncertainty quantification
  • Detection of leakage, bias, censoring, or misleading relationships

A valid result may be that the data does not contain enough reliable signal to justify a model. Knowing that can save time and money.

Packages cover one clearly defined modeling question. Larger datasets, cloud-scale computation, or extensive modeling can be scoped through a custom Fiverr offer.

Please contact me before ordering if your problem is complex.

Domain:

Machine Learning

Deep Learning

Time Series Analysis

Other

Expertise:

Feature learning

Classification

Churn

Clustering

Programming language:

Python

R

SQL

Julia

Tools:

Jupyter Notebook

OpenCV

TensorFlow

MLflow

Colab

Technology:

Python

Julia

scikit-learn

Jupyter Notebook

SQL

Pandas

Models & methods:

Machine Learning

Supervised learning