ServicesStage 04 · Operate11 / 11

Deployment Support and Capability Development

Keep solutions useful as conditions change

Model performance in serviceweekly evaluation · retrain on threshold
0.800.840.880.920.96week 1week 52retrain threshold 0.88retrainweek 14 · retrained · F1 0.879 → 0.939retrainweek 29 · retrained · F1 0.872 → 0.937retrainweek 41 · retrained · F1 0.881 → 0.930week 1 · F1 0.927week 3 · F1 0.913week 5 · F1 0.911week 7 · F1 0.897week 9 · F1 0.887week 11 · F1 0.887week 13 · F1 0.875week 15 · F1 0.939week 17 · F1 0.923week 19 · F1 0.914week 21 · F1 0.899week 23 · F1 0.894week 25 · F1 0.890week 27 · F1 0.882week 29 · F1 0.872week 31 · F1 0.928week 33 · F1 0.919week 35 · F1 0.911week 37 · F1 0.900week 39 · F1 0.897week 41 · F1 0.881week 43 · F1 0.931week 45 · F1 0.920week 47 · F1 0.911week 49 · F1 0.896week 51 · F1 0.894
F1 scoreweek 51 · F1 0.883 · 3 retrains

Deployment support and capability

Analytical models, connected devices and integrated applications require attention after deployment. Data changes, equipment is modified and operating conditions evolve. MNC Analytics helps organisations monitor these changes and maintain the relevance and reliability of their solutions.

We also support the internal capability needed to operate, interpret and improve the systems we help develop.

What we offer

8

  1. 01

    Deployment planning, release management and version control

  2. 02

    Monitoring of data quality, model performance and device health

  3. 03

    Model retraining, recalibration and controlled updates

  4. 04

    Technical incident investigation and corrective maintenance

  5. 05

    Operational runbooks, support arrangements and periodic reviews

  6. 06

    Data and model governance, access requirements and accountability

  7. 07

    Training for analysts, engineers, operators and management teams

  8. 08

    Documentation, handover and practical knowledge transfer

What this enables

A sustainable operating model with clear responsibilities, documented procedures and a process for responding when performance changes.

Related work

All work
01Manufacturing and industrial operationsSpain

Predicting drive failures on a cement plant’s mills and kiln

Vibration and motor-current sensing on gearboxes and main drives, fused with process data, gave maintenance teams three to six weeks of warning before bearing and gear faults.

  • 3–6 wkwarning before confirmed faults
  • -71%unplanned drive stoppages, year on year
  • 0missed faults in the pilot year
Read more
Unplanned drive stoppage hours per quarterBeforeAfter
-50050100150200Q1Q2Q3Q4Q1 Before: 62 hoursQ2 Before: 0 hoursQ3 Before: 118 hoursQ4 Before: 41 hoursQ1 After: 18 hoursQ2 After: 0 hoursQ3 After: 0 hoursQ4 After: 46 hours
hoursQ4 · 46 hours
02Financial and commercial operationsGermany

Detecting billing anomalies in a services business

Contract, work-order and invoice data were reconciled, and an anomaly model now flags unbilled work and pricing errors for review each week, with a governance process for the model itself.

  • 1.9%of revenue identified as leakage
  • 73%of flagged items confirmed by analysts
  • Weeklyreview cycle owned by the finance team
Read more
Flagged items confirmed per weekConfirmed
020406080W1W2W3W4W5W6W7W8W9W10W11W12W1 Confirmed: 31 itemsW2 Confirmed: 44 itemsW3 Confirmed: 52 itemsW4 Confirmed: 48 itemsW5 Confirmed: 39 itemsW6 Confirmed: 35 itemsW7 Confirmed: 30 itemsW8 Confirmed: 27 itemsW9 Confirmed: 24 itemsW10 Confirmed: 22 itemsW11 Confirmed: 21 itemsW12 Confirmed: 19 items
itemsW12 · 19 items
Next service · 01Data Science and Artificial Intelligence

Contact

Tell us about the problem you need to solve

You may have a defined project, an underperforming process or a technical question that needs investigation. We can help assess the challenge, identify the information required and define an appropriate next step.

Discuss your projecthq@mncanalytics.com