RESEARCH / MORTGAGE ANALYTICSAggregate data only

MULTI-AGENT DATA ANALYST

A clearer view of mortgage data.

Explore the supplied dataset, compare segments, and trace every insight to its evidence.

Loading the audited dataset…
Whole dataset

SUPPLIED LOAN CAPACITY

Average maximum loan by area

USD

Dataset-provided maximum amounts; these are not lending offers.

EXISTING OBLIGATIONS

Debt-to-income distribution

Records

Monthly existing debt ÷ monthly income. New mortgage payments are excluded.

Segment explorer

Compare area and credit score cohorts within your selected filters.

Mortgage cohort summary
AreaCredit scoreRecordsAvg. incomeAvg. max loanExisting DTI

STRUCTURED EXECUTION

Agents with a traceable workflow

REPORTING AGENT / COMPUTED BRIEF

What the data tells us

    This brief uses deterministic calculations. Optional AI narration is configured through the Python pipeline; no model is connected to this public dashboard.

    RELATIONSHIP AGENT / WHOLE DATASET

    Associations with maximum loan

    Pearson correlation ranges from −1 to +1. Associations are descriptive, not causal, and do not change with cohort filters.

    Data quality, made visible

    Review flags retained

    Employment-duration review

    No flagged row was silently corrected or removed. The age-18 starting assumption is a review heuristic, not evidence that a record is wrong.

    Preprocessing policy

    Original source values remain in the local archive. Cleaned rows and rejection diagnostics are generated locally; this public dashboard contains cohort statistics only.

    Methods & limitations

    From source to report

    1. Validate 14 required columns, numeric ranges, finite values, and whole-number fields.
    2. Remove exact duplicate records; reject invalid required fields instead of filling values without evidence.
    3. Derive existing DTI and supplied loan-to-income ratio; flag employment duration for review.
    4. Run segment analysis and relationship agents in parallel, then assemble the report.
    Reproduce the analysis

    Read these results in context

    The file supplies a maximum loan amount but does not explain how it was calculated. Collection methods, authenticity, and its licence have not been verified.

    No default, approval, repayment-performance, loan term, or date fields are available. This project does not estimate default risk, make eligibility decisions, or claim forecast accuracy.

    Gender, marital status, education, and age are retained locally for provenance. They are not used to make lending recommendations.

    Dataset source page
    Source fingerprint & reproducibility

    Python 3.10+ · Standard library · No paid API required for analytics · Optional local Ollama or OpenAI-compatible narration.