What analytical methods does AI use?

Fishbone (Ishikawa) diagram — AI automatically categorises causes into 6M (Man, Machine, Method, Material, Measurement, Mother Nature). 5 Whys — AI reconstructs the causal chain from historical data. Pareto analysis — identifies the 20 % of causes responsible for 80 % of problems.

What data does AI correlate?

Production parameters (temperature, speed, pressure) at the time of defect, tool or raw material changes, work shift and operator, measurement results, weather conditions (for thermal expansion), machine maintenance history.

Can AI identify a cause that humans would miss?

Yes — typically correlations that humans lack the capacity to calculate manually: e.g. defect rate rises with the combination (temperature >28 °C AND humidity >65 % AND machine CNC-3), but not with any single factor alone.

AI root cause analysis: systemic solutions instead of firefighting

When the same defect appears for the third time, it is not bad luck — it is a systemic problem that was not resolved at the first occurrence. AI root cause analysis (RCA) helps find the true cause, not just the symptom.

Why traditional RCA fails

Manual RCA has three systemic weaknesses:

Selective data — the analysis is done by someone who only knows what they observed
Confirmation bias — the team looks for the cause it already suspects
One-dimensional analysis — factors are examined in isolation, not in combination

AI correlates all available data at once and searches for statistical patterns — without bias.

How AI RCA works

Step 1: Collecting incident data

For every defect or complaint AI automatically pulls the context:

Production parameters at the time of defective parts
Active tools, their service life, last replacement
Raw materials used and material batch
Work shift and operator
Machine condition and last maintenance
External factors (temperature, humidity)

Step 2: Correlation with history

Analysis: Surface defect — colour deviation (23 cases in 6 months)

Correlations identified by AI:
  ✓ 91 % of cases: morning shift (06:00–14:00)
  ✓ 87 % of cases: outdoor temperature >25 °C
  ✓ 78 % of cases: material batch from supplier B (not supplier A)
  ✓ Combination temperature >25 °C + supplier B: 96 % of defects

  Unrelated factors: operator, day of week, air humidity

AI conclusion:
  Root cause: Material from supplier B has lower thermal pigment stability.
  Colour deviations occur at temperatures above 25 °C.
  
  Recommended action: Change process parameter (reduce process temp by 5 °C)
  OR switch exclusively to supplier A for this material type.

Step 3: Fishbone diagram (auto-generated)

AI populates the Ishikawa diagram with categories:

Machine: CNC-2 — worn temperature regulator (+12 % of cases)
Material: Supplier B — lower thermal stability (primary cause)
Method: No prescribed material acclimatisation before processing
Man: Morning shift — material cold from overnight storage, no acclimatisation
Measurement: Temperature sensor last calibrated 14 months ago

Step 4: Pareto — where you save the most

Of the 23 cases, 19 (83 %) can be eliminated with a single measure: mandatory 2-hour material acclimatisation before processing when outdoor temperature exceeds 20 °C.

Cost of measure: update an instruction + 2 hours of waiting. Saving: 83 % of colour deviation cases.

AI root cause analysis — why defects occur, not just where