Data Driven Project.

If you have any questions or suggestions for the GUI, please reach out to Peter Krogstrup on teams or in the hallway :)

The Project Objective

Objective type:

The Metric of Interest (MOI):

MOI objective:

Second Metric of Interest (MOI2):

MOI2 objective:

Factor type:

Shared project data:

No shared data file loaded.

Select the project stage below

Parameter correlations

No parameter data loaded.

Ranked linear correlations

Rank	Output	Output metadata	Class	Input parameter	Units	Protocol	User	Valid pairs	Correlation (r)	P-value

Correlation matrix

Comparison Correlation matrix

Scatter plots

Y-axis: x-axis: Fit:

Series	x user	x protocol	Y user	Y protocol	Pairs	r

Stage 1 | Exploratory

Stay mission-driven with exploration

There are many possible routes to improve a project's starting point for a given objective - new materials, new device designs, new fabrication methods, new tool designs, and more. The key is choosing the right directions to explore first.

Exploration Prioritization Guidance

1) Describe the potential mission impact clearly.
Explain what new capability or performance window the direction could open relative to the project objective, and use the Pathfinder framework to assess mission relevance.

2) Prioritize by quantitative value.
Evaluate timeline and expected impact in MOIs (Quality, Speed, Scale, OpEx, CapEx) and expected impact on device functionality.

Open Pathfinder Framework: https://the-pathfinder-framework.qfcph.com/

Stage 2 | Reproducibility

Identify identical factor value sets

Reproducibility scope:

No parameter data loaded.

Group	Matching factors	Group subjects	Objective MOI	Valid n	Mean	Sample sigma	CV (%)	Range	Target evaluation

Variation analysis
Process control verification to confirm stable output and repeatability before screening.

Reproducibility Setup

Target for Subject ID-to-Subject ID variability (%). Note: Aim for lower variability than expected effect sizes in Stage 3. Target for Subject ID-to-Subject ID variability for MOI2 (%).

Percent variability lines are shown in the plot together with sigma lines.

Number of samples with identical sets of PCPs (n):

Number of PMPs per subject ID:

If your file has the same column names for MOIs and PMPs as specified here, it will locate them and import automatically.

MOI (mean) values from all n Subject IDs

MOI2 values from all n Subject IDs

Enter or load values, then click 'Plot Reproducibility'.

MOI vs PMP linear correlations

Correlation coefficient (r) describes the strength and direction of the linear relationship between MOI and PMP, from -1 to +1.
P-value (p) measures evidence against no linear correlation; a smaller p-value indicates stronger evidence of a relationship.
Add PMP values and click 'Plot results' to calculate correlations.

Variability troubleshooting

Understand the Sources of Variability

Run-to-run variability can arise from several sources. Here are some suggestions to consider:

Equipment drift (e.g., temperature fluctuations, calibration shifts). Check PMPs.
Material batch differences? (e.g., substrate properties, chemical purity)
Operator differences (e.g., manual handling, setup inconsistencies). Automize and standardize.
Environmental factors (e.g., humidity, vibrations, electromagnetic interference). Log these as PMPs.
Measurement noise (e.g., instrument precision, readout errors)

Process CV vs objective MOI

Process	CV% (PCPs, subjects)

Load Stage 2 data to calculate process CVs.

Assess existing screening data

No screening data loaded.

PB design	Factors	Other factors varying	Objective MOI	Reusable / required runs	New runs	Assessment
No PB completion candidates available.

Experiment Setup

Number of factors (PCPs):

Screening DOE type:

Set PCP count and names before suggesting DOE.

Process Control Parameters (PCPs)

Click 'Set PCP Count' to generate PCP name, low, high, and units fields.

Ready

After experiments

Generate DOE first, then enter MOI Mean (and optional Spread) for each Subject ID.

Effect ranking output will appear here after entering Subject ID MOI values.

MOI2 effect ranking

Effect ranking output will appear here after entering Subject ID MOI2 values.

Transfer key factors to Stage 4

Cumulative effect threshold (%): Ranking:

Stage 4 | Optimization

After identifying the most critical PCPs with the largest effect on the MOI in Stage 3, optimize the MOI using Response Surface Methodology (RSM) with a quadratic model. Use Central Composite Design (CCD) to capture the true optimum efficiently.

Use Stage 3 effect ranking to choose key PCPs, then set low/high levels for CCD planning (center is the midpoint).

Typical CCD Subject ID count: 2^k + 2*k + center points, where k is the number of key PCPs. Enter the shortlisted key PCPs and their low/high optimization bounds below.

Optimization Setup

Planning mode:

Recommend a DoE Design manually

Optimization method:

Number of key PCPs (k):

Center points:

Levels per factor:

CSV/TSV/XLSX: one DP per row; first column is its name, remaining columns are numeric levels (optional header: DP, Level 1, ...).

Maximum Subject IDs (required):

PCP execution:

Optimization factors

Click 'Set key PCP count' to generate optimization factor name, low, high, and units fields.

Generate the fixed RSM/CCD optimization matrix, including factorial, axial, and center-point runs.

After executing the basic optimization DOE add adaptive axial points here for finetuning optimum.

Enter factor bounds, then click 'Generate Optimization Subject IDs'.

After experiments

Generate suggested Fixed RSM (CCD) Subject IDs first. Then enter MOI mean for each Subject ID.

Optimization Visualization