Perform a Deep Confusion Matrix Analysis to Diagnose Model Errors

Act as an expert in machine learning model diagnostics. Analyze the following confusion matrix and provide a comprehensive error diagnosis. **Confusion Matrix:** ``` [PASTE_CONFUSION_MATRIX — e.g., a 2D array or formatted table with actual vs. predicted labels] ``` **Class Labels (in order):** [LIST_CLASS_LABELS] **Problem Context:** - Task description: [TASK_DESCRIPTION] - Model used: [MODEL_NAME] - Total dataset size: [DATASET_SIZE] - Known data quality issues (if any): [DATA_ISSUES] **Please deliver the following analysis:** 1. **Matrix interpretation:** Walk through the confusion matrix cell by cell, explaining what each value means in plain language using the actual class names from [LIST_CLASS_LABELS]. Highlight true positives, true negatives, false positives, and false negatives for each class. 2. **Derived metrics calculation:** From the matrix, manually compute and present: accuracy, precision, recall, specificity, and F1-score for each class. Show the formulas used. 3. **Error pattern identification:** Identify the most common misclassification pairs. For each pair, explain which class is being confused with which and hypothesize why this confusion might occur in the context of [TASK_DESCRIPTION]. 4. **Class imbalance impact:** Assess whether class imbalance is inflating or deflating overall accuracy. Calculate and compare accuracy vs. balanced accuracy. 5. **Visualization recommendations:** Suggest 2-3 visualization techniques (e.g., normalized confusion matrix, error heatmap, per-class bar charts) and explain what each would reveal. 6. **Targeted improvement plan:** For the top 3 most problematic misclassification pairs, provide specific remediation strategies including: feature engineering ideas, data collection suggestions, model architecture changes, or post-processing rules. 7. **Threshold analysis (for binary):** If binary classification, recommend how adjusting the decision threshold would shift the matrix and which direction to adjust based on the problem's cost structure. Format with clear sections, tables for computed metrics, and prioritized recommendations.