# Technical Data Extraction: Accuracy vs. Max Cluster Size
## 1. Chart Overview
This image is a grouped bar chart comparing the performance (Accuracy) of two categories, **Coding** and **MC**, across different **Max cluster sizes**.
## 2. Axis Information
* **Y-Axis Label:** Accuracy (%)
* **Y-Axis Scale:** 0.0 to 0.4 (with markers at 0.0, 0.1, 0.2, 0.3, 0.4)
* **X-Axis Label:** Max cluster size
* **X-Axis Categories:** None, 32, 16, 8, 4, 2
## 3. Legend
* **Blue Bar:** Coding
* **Pink Bar:** MC
## 4. Data Table Extraction
The following table represents the approximate values extracted from the bar heights:
| Max cluster size | Coding (Blue) Accuracy | MC (Pink) Accuracy |
| :--- | :--- | :--- |
| **None** | ~0.41 | ~0.45 |
| **32** | ~0.12 | ~0.45 |
| **16** | ~0.11 | ~0.45 |
| **8** | ~0.11 | ~0.45 |
| **4** | ~0.21 | ~0.44 |
| **2** | ~0.00* | ~0.36 |
*\*Note: For the "2" cluster size, the Coding bar is not visible, indicating a value of 0 or near 0.*
## 5. Key Trends and Observations
* **MC Performance Stability:** The "MC" category maintains a very high and stable accuracy (approx. 45%) for cluster sizes "None" through "8". There is a slight dip at size "4" and a significant drop to approximately 36% at size "2".
* **Coding Performance Volatility:** The "Coding" category performs best when the Max cluster size is "None" (~41%).
* **Impact of Clustering on Coding:** Introducing a cluster size (32, 16, or 8) causes a sharp decline in Coding accuracy to roughly 11-12%.
* **Recovery at Size 4:** Interestingly, Coding accuracy sees a relative recovery at cluster size "4", rising back to approximately 21%, before disappearing at size "2".
* **Comparative Gap:** Except for the "None" category where the two are relatively close, the "MC" category significantly outperforms the "Coding" category across all specific cluster size constraints.
