03 // Interactive Compute

Machine Learning & Mathematical Lab

Zero-dependency, pure client-side engineering instruments. Real-time statistical analysis, canvas gradient visualizers, and dataset profiling operating directly within your browser runtime.

Tool 1: Confusion Matrix & Diagnostic Metric Analyzer

[Pure ES6+ / Real-time Math]

Raw Class Distribution Inputs

True Positives (TP)

False Positives (FP)

False Negatives (FN)

True Negatives (TN)

Modifying any quadrant updates the proportion heatmap, accuracy, precision, sensitivity, specificity, and harmonic F1-score immediately with zero server latency.

Predicted Class: Positive vs. Negative

Actual Pos / Pred Pos (TP) 86 43.0%

Actual Pos / Pred Neg (FN) 13 6.5%

Actual Neg / Pred Pos (FP) 9 4.5%

Actual Neg / Pred Neg (TN) 92 46.0%

Accuracy

89.0%

(TP + TN) / Total

Precision (PPV)

90.5%

TP / (TP + FP)

Recall (TPR)

86.9%

TP / (TP + FN)

Specificity (TNR)

91.1%

TN / (TN + FP)

F1-Score

88.7%

2·(P·R) / (P + R)

Tool 2: Activation Function & Gradient Visualizer

[HTML5 Retina Canvas / Dynamic Derivatives]

Activation Function:

Scalar Input (x): 0.00 Slide thumb or drag directly across canvas curve

Formula f(x): f(x) = max(0, x)

Derivative f'(x): f'(x) = 1 if x > 0 else 0

Activation f(x): 0.0000

Gradient f'(x): 0.0000

Tool 3: In-Browser CSV & Dataset Profiler

[100% Client-Side / Zero Server Upload]

⇪

Drag & Drop any .CSV dataset here

or click to browse your local filesystem (All data remains private in memory)

Total Records (Rows): 10
Feature Attributes (Cols): 8
Null / Missing Values: 1
Memory Footprint Est.: 640 B

Statistical Schema & Feature Profiles

Column Attribute	Detected Type	Missing Count	Minimum	Maximum	Mean (μ)	Std Dev (σ)

Data Stream Sample Preview (First 8 Records)