Video summary

Learning Curve Analysis (Incremental Unit Time Model, Developing & Explaining Equations & Graphing)

Main summary

Key takeaways

Educational

Main ideas / lessons conveyed

  • Learning curve (experience curve) concept: As production quantity increases, efficiency improves, so individual unit time/cost decreases.
  • Focus on the “incremental time model” / “Crawfords model”: Rather than modeling only individual unit time, it computes total (cumulative) time/cost by using an algebraic midpoint of a production lot.
  • Key parameters:
    • Learning rate (e.g., 80% learning): Interpreted as: when output doubles, unit time/cost decreases by 20%.
    • Improvement rate: Complementary to learning rate (here, 20% improvement rate).
  • Graph interpretation:
    • Incremental unit time graph:
      • y-axis = individual unit time/cost
      • x-axis = number of units produced
    • Total cumulative time graph:
      • y-axis = total time/cost
      • x-axis = lot size / number of units considered

Methodology / equations / steps (detailed)

1) Define the learning curve basics

Use a learning curve characterized by:

  • Learning rate = 80% (meaning unit time/cost decreases by 20% when production doubles)
  • Improvement rate = 20% (implied complement)

Example “base” value:

  • First unit time/cost: (a = 100)
    • Interpretation: (100) represents direct labor hours for the first unit.

2) Individual unit time model (incremental learning curve)

Equation for individual unit time:

[ y = a \, x^{b} ]

where:

  • (y) = individual unit time/cost
  • (a) = cost/time of the first unit (given as 100)
  • (x) = number of units produced
  • (b) = learning exponent

Compute the exponent (b) from the learning rate:

  • For an 80% learning curve:

[ b = \frac{\log(0.80)}{\log(2)} ]

  • Approximation:

[ b \approx -0.322 ]

Meaning of the model (80% learning):

  • When output doubles, unit time/cost becomes 20% less, so the curve decreases as (x) increases.

Illustrative individual times mentioned:

  • For (x=2): individual time/cost (\approx 80)
  • For (x=4): (\approx 64)
  • For (x=8): (\approx 51)

(These values are used to demonstrate the decreasing pattern.)

3) Total (cumulative) time via summing individual unit times

Core idea:

  • Total time/cost for producing up to (x) units is the sum of incremental (individual) unit times.

Summation concept:

  • Sum (y) values from unit 1 through unit (x).

Example cumulative totals described:

  • After 1 unit: (100)
  • After 2 units: (100 + 80 = 180)
  • After 3 units: (180 + 70 = 250)
  • After 4 units: (250 + 64 = 314)

Key takeaway:

  • This approach is incremental and cumulative: each added unit contributes its own incremental time.

4) Crawfords (Incremental Unit Time) model for total time/cost using a lot midpoint

4a) Incremental unit time at the lot midpoint

Model equation (as described):

[ y = a \, K^{b} ]

where:

  • (y) = midpoint incremental time/cost for the lot
  • (a) = first-unit time/cost (100)
  • (K) = algebraic midpoint of the lot
  • (b) = learning exponent

Meaning of (K):

  • (K) is not a constant rate; it is computed from the lot’s unit positions (depends on lot size).

4b) Total time/cost using lot size and midpoint incremental time

Total time/cost is computed as:

[ \text{Total} = x \cdot y ]

  • Subtitles also express it (depending on substitution) as:

[ x \cdot a \cdot K^{b} ]

Interpretation (from subtitles):

  • (x) = lot size / number of units being produced
  • (y) = incremental time at midpoint (K)
  • Multiply to get total time/cost for the lot.

5) How to compute (K) (algebraic midpoint of a lot) — detailed equation

Inputs:

  • (N_1) = index of the first unit in the lot
  • (N_2) = index of the last unit in the lot
  • (L) = number of units in the lot (so the span runs from (N_1) to (N_2))
  • (b) = learning exponent

Conceptual structure of the midpoint formula:

  • (K) is computed via an algebraic midpoint expression involving terms like:
    • differences of ((N^{1+b}))-type quantities
  • Then a normalization is applied using (L) and exponent manipulation (the narration conveys the idea even if formatting is unclear).

Subtitles’ stated dependency (simplified idea):

  • (K) depends on:
    • ((N_1 + L - 1/2)^{1+b} - (N_1 - 1/2)^{1+b})
  • with additional normalization and exponent-handling that yields the final (K).

5a) Example of (N_1), (N_2)

Example lot:

  • Lot has 4 units
  • Units are 3, 4, 5, 6

So:

  • (N_1 = 3)
  • (N_2 = 6)

Midpoint offsets:

  • (N_1 - 1/2 = 2.5)
  • (N_2 + 1/2 = 6.5)

6) Final graph interpretation (cumulative total time vs lot size)

The speaker contrasts:

  • A curve for a different learning rate (90% learning, shown as green)
  • The primary curve of interest: 80% learning (red)

Key explanation:

  • For cumulative total time at increasing lot sizes, use:
    • the lot midpoint (K)
    • the learning exponent (b)
    • and the total expression based on (x \cdot y)

Example cumulative totals mentioned:

  • The subtitles convey that cumulative total time/cost grows to large values (e.g., an ending cumulative total like “892” after summing to a stated output count).

Speakers / sources featured

  • No specific named speakers are identified in the subtitles.
  • The only referenced “source” concept is the Crawfords model (Crawford’s model) for the incremental unit time / learning approach.

Original video