Modeler
Modeler. AutoML built for production. White-box by design.
TIMi’s predictive modeling engine. Tests 15,000+ model configurations in seconds. Delivers a stable, fully explainable model. Ready for production in one drag & drop. Built on +30 years of data science expertise, automated into a single click.
How it works
One click. 15,000 models evaluated. The best one delivered.
Modeler lives inside the Anatella pipeline. No export. No handoff. Predictive modeling starts where data preparation ends, in the same graph, on the same infrastructure.
What the analyst sees
- One Click AutoML — drag & drop inside the Anatella graph
- Results in seconds — no configuration required
- Word and PNG reports generated automatically
- No modeling expertise required to get started
- Full manual override available at any step
What runs underneath
- 15,000+ model configurations evaluated per run
- Algorithms : Ridge · Lasso · Neural Net · Forest · Trees
- 100% multithreaded · vectorized · cache-optimized engine
- 4 anti-overfitting layers: 99% of cases overfitting-free
- Drift control: unique to TIMi
No handoff with Anatella
Modeler nodes appear directly in the Anatella graph. Data prepared in Anatella feeds Modeler instantly. Same memory space, zero duplication, zero export overhead.
Full automation, full control
Every step is automated by default. Variable selection, meta-parameter optimization, cross-validation, report generation. Switch to manual mode at any point to override any parameter.
Full portability: no lock-in
Export any model to SQL, Python, SAS, VBA, Javascript or HTML in one click. Native integration into any existing stack. Full independence from TIMi infrastructure if needed.
Data consolidation and feature engineering
Merge heterogeneous sources (ERP, CRM, CSV, APIs, databases) into a single clean, structured dataset, ready for reporting or modeling. The universal first project.
Used by MTN to consolidate 12.1B records from multiple sources into a single churn model.
Automated reporting
Automatically update MS Office reports, Power BI, Tableau, Qlik and Kibella dashboards at recurring intervals. Set it once, run it forever. No manual intervention.
Data anonymization
Anonymize, pseudonymize and mask sensitive data across datasets of several billion rows. Fully auditable. Built for regulated industries where compliance is not optional.
Required by 100% of EU-regulated organizations under GDPR Article 25
Feature engineering for machine learning
Prepare variables for Modeler : transformations, encoding, missing value handling, complex KPI creation. Anatella as the native preparation layer before AutoML, inside the same graph.
Used by BNP Paribas to prepare 24,000 variables for a 98% accuracy credit model.
Social Network Analysis and Graph Mining
Run SNA and graph mining algorithms on graphs with tens of millions of nodes and billions of arcs. Compute communities, social leaders, fraud networks, in a no-code interface.
Performance and Capabilities
Every modeling need. Automated or manual. At any scale.
Speed and scale are architectural properties of Modeler, not configuration options. Every capability below is automated by default and fully overridable manually.
+ 15.000
model configurations evaluated per run · 100% multithreaded · vectorized · cache-optimized
+ 30.000
variables supported · billions of rows · no infrastructure overhead
1 GB
dataset stored in under 50 MB RAM · proprietary compression engine
2.4M
records/second · batch scoring via Anatella integration
Where a team builds a dozen models a year, Modeler enables those same analysts to build hundreds of thousands — and become genuine business experts in their organization.
Accuracy and modeling
Top used
Robustness and governance
Deployment and integration
Most Common Uses
One engine. Every prediction problem.
From credit scoring to predictive maintenance, five use cases that illustrate what Modeler makes possible across industries, at scale.
Any sources
Any destination
Any infrastructure
On-premiseTop used
Runs on your servers. Your data center. Your hardware. Full control, zero external dependency. Installs in under 1 minute.
Private cloudTop used
Deploy on a private cloud you control. Same engine, same performance, no shared tenancy.
Edge
Run next to the data source. Process locally with no round-trip to a distant data center.
Isolated deployment
Air-gapped environments. No outbound connection required. Full operational independence.
Hybrid
Mix on-premise, cloud and edge. One platform, one language, any topology.
Unique to TIMi
Your model was accurate on day one. Is it still valid today?
Most AutoML platforms tell you how accurate your model is at training time. The Drift Report tells you when it starts to expire, automatically, continuously, in production.
What drift is
A predictive model is only valid as long as the data patterns it learned still hold. When customer behavior shifts, when market conditions change, when new fraud patterns emerge, your model drifts. Silently. Without warning. Until it produces unreliable outputs.
What the Drift Report does
Modeler continuously evaluates model validity after deployment. The Drift Report is available at any time, giving data teams a clear, up-to-date view of whether each model still reflects current data patterns. Know before it fails.
Why it matters in regulated environments
In regulated industries, a drifting model is not only a performance issue, it is a compliance risk. The Drift Report provides a continuous audit trail proving that models were monitored, flagged and updated. AI Act aligned by design.
Included in every Modeler output by default. Found nowhere else.
Client Results
Unlock the full value of your data without compromise.
Discover how our users put TIMi to work every day.
+ 21% net margin
ING – Banking
“TIMi’s model identified the top 15% of business clients most likely to need credits – at exactly the right moment in the year. 26% of those contacted bought a product. Result: €12.3M net margin, €1.3M cost reduction.”
X 8,5 conversion
BNP Paribas – Banking
“TIMi’s model identified the top 15% of business clients most likely to need credits – at exactly the right moment in the year. 26% of those contacted bought a product. Result: €12.3M net margin, €1.3M cost reduction.”
+ 250M investable
AXA – Insurance
“TIMi’s model identified the top 15% of business clients most likely to need credits – at exactly the right moment in the year. 26% of those contacted bought a product. Result: €12.3M net margin, €1.3M cost reduction.”
Get in touch
Stop renting compute power. Start owning your company.
Discuss your constraints, your volumes, your governance requirements.
We will show you how TIMi fits into your architecture.
